U.S. Department of Commerce Volume 102 Number 1 January 2004 Fishery Bulletin U.S. Department of Commerce Donaid L. Evans Secretary National Oceanic and Atmospheric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fisheries .^TOFCo. X K 1 ^ / The Fishery' Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- N( >AA, 7600 Sand Point Way NE, BIN C15700, Seattle. WA 981 15-0070. Periodicals postage is paid at Seattle, WA, and at additional mailing offices. POST- MASTER: Send address changes for sub- scription- to Fishery Bulletin. Superin- tendent of Documents, Attn.: Chief. .Mail List Branch, Mail Stop SSOM, Washing- ton. DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Set if Commerce has deter- mined that the publication of tin ording to law for the transaction of public business of this Department. Use of funds for printing of nodical has been approved by the oroftheOffii cement and Budget. For sale by the Superintendent of nuts. US. Government Printing I mice, Washington, DC 20402. Subscrip- tion pi i it: $55.00 domestic and $68.75 foreign. Cost per single issue: $28.00 dome ,5.00 foreign. See back for order form. Scientific Editor Dr. Norman Bartoo Associate Editor Sarah Shoffler National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 981 15-0070 Editorial Committee Dr. Harlyn O. Halvorson Dr. Ronald W. Hardy Dr. Richard D. Methot Dr. Theodore W. Pietsch Dr. Joseph E. Powers Dr. Harald Rosenthal Dr. Fredric M. Serchuk Dr. George Watters University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service National Marine Fisheries Service Fishery Bulletin web site: www.fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery scien ring, and economics. It began as the Bulletin of the United States Pish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents tl volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 196.1 ired as a numbered bulletin. A new system began in 1963 with volume 6:3 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70. number 1. January 1972, the Fishery Bulletin became a lieal, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents. U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libl irch institutions. State and Federal agencies, and in exi for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 102 Number 1 January 2004 Fishery Bulletin Contents ary MAR 5 2004 The conclusions and opinions expressed in Fisher)' Bulletin are solely those of the authors and do not represent the official position of the National Marine Fisher- ies Service (NOAA) or any other agency or institution. The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprietary product or pro- prietary material mentioned in this pub- lication. No reference shall be made to NMFS. or to this publication furnished by NMFS, in any advertising or sales pro- motion which would indicate or imply that NMFS approves, recommends, or endorses any proprietary product or pro- prietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the advertised product to be used or purchased because of this NMFS publication. Articles 1-13 Alonzo, Suzanne H., and Marc Mangel The effects of size-selective fisheries on the stock dynamics of and sperm limitation in sex-changing fish 14-24 Baba, Katsuhisa, Toshifumi Kawajiri, Yasuhiro Kuwahara, and Shigeru Nakao An environmentally based growth model that uses finite difference calculus with maximum likelihood method: its application to the brackish water bivalve Corbicula /aponica in Lake Abashiri, Japan 25-46 Brodeur, Rick D., Joseph P. Fisher, David J. Teel, Robert L. Emmett, Edmundo Casillas, and Todd W. Miller Juvenile salmomd distribution, growth, condition, origin, and environmental and species associations in the Northern California Current 47-62 Garcia-Rodrfguez, Francisco J., and David Aurioles-Gamboa Spatial and temporal variation in the diet of the California sea lion (Zalophus californianus) in the Gulf of California, Mexico 63-77 Jung, Sukgeun, and Edward D. Houde Recruitment and spawning-stock biomass distribution of bay anchovy (Anchoa mitchilli) in Chesapeake Bay 78-93 Kellison, Todd G., and David B. Eggleston Coupling ecology and economy: modeling optimal release scenarios for summer flounder (Paralichthys dentatus) stock enhancement 94-107 Kritzer, Jacob P. Sex-specific growth and mortality, spawning season, and female maturation of the stripey bass (Lut/anus carponotatus) on the Great Barrrier Reef Fishery Bulletin 102(1) 108-117 Orr, Anthony J., Adria S. Banks, Steve Mellman, Harriet R. Huber, Robert L. DeLong, and Robin F. Brown Examination of the foraging habits of Pacific harbor seal (Phoca vitulina richardsi) to describe their use of the Umpqua River, Oregon, and their predation on salmonids Companion paper with Purcell et al., see "Notes" below. 118-126 Park, Wongyu, R. Ian Perry, and Sung Yun Hong Larval development of the sidestriped shrimp (Pandalopsis dispar Rathbun) (Crustacea, Decapoda, Pandahdae) reared in the laboratory 127-141 Pearson, Donald E., and Franklin R. Shaw Sources of age determination errors for sablefish (Anop/opoma fimbria) 142-155 Powell, Allyn B., Robin T. Cheshire, Elisabeth H. Laban, James Colvocoresses, Patrick O Donnell, and Marie Davidian Growth, mortality, and hatchdate distributions of larval and juvenile spotted seatrout (Cynoscion nebulosus) in Florida Bay, Everglades National Park 156-167 Santana, Francisco M., and Rosangela Lessa Age determination and growth of the night shark (Carcharhinus signatus) off the northeastern Brazilian coast 168-178 Smith, Keith R„ David A. Somerton, Mei-Sun Yang, and Daniel G. Nichol Distribution and biology of prowfish (Zaprora silenus) in the northeast Pacific 179-195 Ward, Peter, Ransom A. Myers, and Wade Blanchard Fish lost at sea: the effect of soak time on pelagic longlme catches 196-206 Watanabe, Chikako, and Akihiko Yatsu Effects of density-dependence and sea surface temperature on interannual variation in length-at-age of chub mackerel (Scomber japonicus) in the Kuroshio-Oyashio area during 1970-1997 Notes 207-212 Llanos-Rivera, Alejandra, and Leonardo R. Castro Latitudinal and seasonal egg-size variation of the anchoveta (Engrauhs nngens) off the Chilean coast 213-220 Purcell, Maureen, Greg Mackey, Eric LaHood, Harriet Huber, and Linda Park Molecular methods for the genetic identification of salmonid prey from Pacific harbor seal (.Phoca vitulina richardsi) scat Companion paper with Orr et al., see "Articles" above. 221-229 Weng, Kevin C, and Barbara A. Block Diel vertical migration of the bigeye thresher shark (Alopias superciliosus), a species possessing orbital retia mirabilia 231 Subscription form Abstract— Fisheries models have tradi- tionally focused on patterns of growth, fecundity, and survival offish. However, reproductive rates are the outcome of a variety of interconnected factors such as life-history strategies, mating patterns, population sex ratio, social interactions, and individual fecundity and fertility. Behaviorally appropriate models are necessary to understand stock dynamics and predict the success of management strategies. Protogynous sex-changing fish present a challenge for management because size-selective fisheries can drastically reduce repro- ductive rates. We present a general framework using an individual-based simulation model to determine the effect, of life-history pattern, sperm production, mating system, and man- agement strategy on stock dynamics. We apply this general approach to the specific question of how size-selective fisheries that remove mainly males will impact the stock dynamics of a protogynous population with fixed sex change compared to an otherwise identical dioecious population. In this dioecious population, we kept all aspects of the stock constant except for the pattern of sex determination (i.e. whether the species changes sex or is dioecious). Protogynous stocks with fixed sex change are predicted to be very sensitive to the size-selective fishing pattern. If all male size classes are fished, protogynous populations are predicted to crash even at relatively low fishing mortality. When some male size classes escape fishing, we predict that the mean population size of sex-chang- ing stocks will decrease proportionally less than the mean population size of dioecious species experiencing the same fishing mortality. For protogynous spe- cies, spawning-per-recruit measures that ignore fertilization rates are not good indicators of the impact of fishing on the population. Decreased mating aggregation size is predicted to lead to an increased effect of sperm limitation at constant fishing mortality and effort. Marine protected areas have the poten- tial to mitigate some effects of fishing on sperm limitation in sex-changing populations. Manuscript approved for publication 23 July 2003 by Scientific Editor. Manuscript received 20 October 2003 at NMFS Scientific Publications Office. Fish. Bull 102:1-13(2004). The effects of size-selective fisheries on the stock dynamics of and sperm limitation in sex-changing fish Suzanne H. Aionzo Institute of Marine Sciences and the Center lor Stock Assessment Research (CSTAR) University of California Santa Cruz 1156 High Street Santa Cruz, California 95064 E-mail address shalonzoiS'ucscedu Marc Mangel Department of Applied Mathematics and Statistics Jack Baskin School of Engineering and the Center for Stock Assessment Research (CSTAR) University of California Santa Cruz 1156 High Street Santa Cruz, California 95064 Fisheries models are generally used to predict the impact of fishing on stock dynamics and yield (Quinn and Deriso, 1999; Haddon, 2001). Classic models have focused mainly on growth, fecundity, and survival of species, with- out considering the impact of mating patterns on reproduction, survival, and recruitment. It is now recognized that life-history strategies and mating behavior will affect stock dynamics. Even so, general quantitative predic- tions regarding the effect of specific life-history patterns on fished popula- tions are limited and further theory is needed (Levin and Grimes. 2002). It is likely that management strategies taking into account a species' reproduc- tive behavior will greatly improve our ability to manage stocks (e.g. Beets and Friedlander, 1999). We would also like to know when the mating behavior and reproductive strategies of a stock will be worth investigating and when tradi- tional management techniques will be sufficient. For example, in a manage- ment context, how do sex-changing stocks differ from separate-sex species? Here, we take an initial step toward generating a theory of the combined effect of life history and mating pat- terns on stock dynamics by focusing on the potential for and effect of sperm limitation in a protogynous (female to male) sex-changing stock. We focus on protogyny for this article because numerous protogynous species are com- mercially important, namely red porgy {Pagrus pagrus), gag grouper iMyc- teroperca microlepis), and California sheephead iSemicossyphus pulcher). Sex-changing fish present a unique challenge for management because size- selective fisheries have the potential to drastically reduce reproductive rates and population size at levels of fishing that would not pose a problem for dioe- cious (separate-sex) species (Huntsman and Schaaf, 1994; Armsworth, 2001; Fu et al., 2001). On the other hand, pro- togynous stocks may be less sensitive to the removal of large individuals if females are not fished and fertilization rates remain high. Many commercially important species are known to change sex (Bannerot et al., 1987; Shapiro, 1987; Coleman et al., 1996; Brule et al., 1999; Adams et al., 2000; Armsworth, 2001; Fu et al., 2001). Previous models have shown that sex-changing fish may be vulnerable to fishing (Bannerot et al., 1987; Huntsman and Schaaf, 1994; Armsworth, 2001; Fu et al.. 2001). Complications arise because the ef- fect of fishing on a sex-changing spe- cies is mediated by many aspects of their reproductive biology, such as sex ratio, size-dependent fecundity, spawn- ing aggregation size, and reproductive skew. Furthermore, patterns of sex change have cascading effects on the sex ratio, social interactions, population Fishery Bulletin 102(1) fecundity, and male sperm production — all of which can affect stock dynamics. Thus, we cannot treat sex change as an isolated aspect of a species. Instead, we must consider sex change within the context of the mating system and the life history of the species to make general predictions. Behaviorally appropriate models are required to gener- ate constructive qualitative and quantitative theory. Past theory has indicated that sex-changing populations exhibit stock dynamics that often differ from those of dioecious populations (Bannerot et al., 1987; Huntsman and Schaaf, 1994; Armsworth, 2001; Fu et al, 2001 ). Furthermore, pro- togynous stocks are predicted to be sensitive to fishing pat- tern and may exhibit nonlinear dynamics that could lead to population crashes (Armsworth, 2001). However, it is not known which aspects of the mating behavior and life his- tory pattern of sex-changing stocks drive these differences. Here we focus on comparing a protogynous stock with an otherwise identical dioecious population to determine the effect of mating aggregation size, fertilization rates, and life history pattern on stock dynamics. Size-selective (or age-selective) fisheries can impact a species through a decrease in spawning stock biomass, in general and through the removal of highly fecund larger and older individuals, in particular (Sadovy, 2001). How- ever, in protogynous species, fisheries that preferentially remove large males can also change the population sex ratio; however, the exact effect of fishing pressure on stock dynamics in a protogynous species is complex. At one extreme, the complete removal of males from the popula- tion would cause a stock to crash, potentially making sex- changing species more vulnerable than dioecious species in the face of high fishing pressures. At the other extreme, sex-changing species may be less affected by size-selective fisheries if female fecundity limits recruitment and males are not removed in such numbers as to reduce mating or fertilization rates. Currently, there is no theory that predicts the potential for sperm limitation in protogynous stocks as a function of gamete production, fertilization rates, and mating pattern. It has been suggested that marine reserves may be a vi- able management option for species where highly fecund older individuals are critical to reproduction (Levin and Grimes, 2002). However, no theory exists that can predict the impact of marine reserves on stock dynamics in sex- changing species. We consider the impact of a no-take marine reserve on the stock dynamics. We compare the effect of setting aside 0-30% of the spawning population in a reserve. We assume that larval production is exported from within the reserve to the rest of the population and determine whether the reserve can mediate some of the ef- fects of fishing outside the reserve because this represents the optimal scenario for marine reserves. We also compare mean catch rates in the presence and absence of a reserve as a function of fishing mortality. Spawning-per-recruit (SPR) measures are often used to estimate the impact of fishing on a stock (Parkes, 2000; Jennings et al., 2001). Ideally, a spawning-per-recruit mea- sure would keep track of per-recruit production of larvae or eggs (Jennings et al., 2001). However, spawning stock biomass per recruit (SSBR) is commonly used to estimate the reproductive output per recruit at different intensities of fishing. One assumes that the biomass of mature fish is linearly related to reproductive output, which may be the case when egg production limits biomass and fecundity in- creases linearly with biomass. In protogynous stocks, over- fishing of males alone may decrease fertilization rates and hence reproductive output without affecting either female biomass or egg production. Thus, in protogynous stocks or sex-selective fisheries, classic measures of spawning per re- cruit may misrepresent the impact of fishing on the stock's reproduction and hence population stability (Punt et al., 1993). We examine a variety of per-recruit measures and determine their ability to predict changes due to exploita- tion in mean population size. In this study, we describe a general approach using sex- and size-dependent individual-based simulation models that predict reproduction, size distribution, and sex ratio in fished populations as a function of mating system and sex-change pattern. We examine the case where sex change occurs at a specific size threshold. We recognize that plastic and socially mediated sex-change patterns have been ob- served, and our results will apply only to species with fixed sex change. We explore the impact of mating aggregation size, sperm production, and asymptotic fertilization rates on the predicted stock dynamics in the presence of exploita- tion. We make predictions regarding the effects of fishing on population size, reproduction, sex ratio, size distribu- tion, and fertilization rates. We also compare our results to previous work and discuss future directions. Methods We used an individual-based simulation to predict the size distribution, individual and population fecundity, popula- tion sex ratio, fertilization rate, and population size as a function of fishing mortality (Fig. 1). Individuals vary in age, size, sex, and mating site. Population size varies as a function of baseline survival, fishing mortality, reproduc- tion, and larval recruitment. Reproduction depends on the pattern of sex change, mating system, sex ratio, mating site, and fecundity (or fertility) of individual males and females. For each annual time period, we determined individual survival, the size and age of these individuals in the next time period, and the total production of surviving offspring by those individuals. Initial analyses showed that a station- ary size, sex, and age distribution is found within approxi- mately 50 time periods and is independent of the initial population conditions. Thus, we simulated 100 time periods prior to examining the impact of fishing on stock dynamics to ensure that the population had already reached the sta- tionary size and sex distribution for that scenario and set of parameters. We then examined the model for 100 repro- ductive seasons in the presence of fishing with a constant mean fishing mortality. Because a number of elements of the model were stochastic, we examined 20 simulations for each scenario and set of parameter values. Initial analyses indicated that 20 simulations were more than sufficient to lead to low variability in the key measures of interest. We assumed that reproduction occurs at the level of the mating Alonzo and Mangel: The effects of size-selective fisheries on the stock dynamics of and sperm limitation in sex-changing fish 3 group at different reproductive sites. Individual sur- vival, maturation, sex change, and mating site were determined stochastically as described below. Fishing and adult survival We assumed that adult survival is density indepen- dent but depends on fishing selectivity, fishing mor- tality, and baseline adult mortality in the absence of fishing. For simplicity, we assumed that age and size do not affect nonfishing adult mortality p. A . We assumed that the fishery is size selective; we let L represent fish size, F represent annual fishing mortality, L f represent the size at which there is 50% chance an individual of that size will be taken, and r represent the steepness of the selectivity pattern. Then fishing selectivity per size class siL) is given by siD- l + exp-HL-L,)) and adult annual survival becomes cr(L) = exp(-/i A -Fs(L)) (1) (2) We assumed that fishing does not differentially affect the sexes independent of size. We recognize, however, that for some species this may not be the case. We also assumed that fishing occurs each year prior to reproduction and can represent either pulse or continuous fishing with an annual mortality F. We let N it) represent the number of individuals in age class a at time t so that population size N(t)= S a N a (t). Population dynamics We assumed that the number of larvae that enter the popu- lation is determined by the production of fertilized eggs Pit) and the probability that those larvae will survive to recruit. Pit) is determined by the adult fecundity and fertilization rates described below. For computational tractability, we also assumed that a population ceiling N max exists (Mangel and Tier, 1993, 1994 ). However, we chose N mBX large enough that the stable population size was below the ceiling. Larval survival has both density-independent and density-depen- dent components (e.g. Cowen et al., 2000; Sale, 2002). We used a Beverton-Holt recruitment function to determine larval survival to the next age class (Quinn and Deriso, 1999; Jennings et al., 2001). Larvae represented the zero- age class N (t) and thus the number of larvae surviving to recruit in any year t is given by N n it) = (oPit))/(l+pPit)) if (ctP(t))/(l+pP(t)) +J j Njt)exp(-A). (4) Mating system We assumed that reproduction occurs at the level of the mating group, and we examined the effect of varying mating group size and the number of mating sites. We assumed Fishery Bulletin 102(1) that juveniles and adults exhibit site fidelity but that larvae settle randomly among mating sites. We also assumed that the population carrying capacity is split equally among the mating sites and that the total capacity of all mating sites exceeds the maximum population size in the absence of fish- ing as determined by adult mortality and the recruitment function. Therefore, mating sites do not limit recruitment but may affect reproductive rates. We examined three cases: 1 ) the entire population mates at one site (one mating site with up to 1000 individuals); 2) a few large mating groups exist ( 10 sites with a maximum of 100 individuals per site); and 3) many small mating aggregations exist (20 mating sites with a maximum of 50 individuals per site). For sim- plicity, we assumed that within a mating site, individuals mate in proportion to their fertility and fecundity. Therefore, large males and females have higher expected reproductive success. However, we assumed that all males that are large enough to change sex have a chance of reproducing propor- tional to their fertility. This is equivalent to assuming that females exhibit a mate choice threshold I Janetos, 1980) that has evolved with the size-at-sex change and that females have an equal probability of mating with males above this size threshold. However, a large male mating advantage clearly still exists. We also assumed that fishing mortality remains constant as mating aggregation size varies. Thus, we assumed that fishing effort per site does not increase as the number of mating sites decreases. An alternative would be to assume that total fishing mortality increases as the number of mating aggregations decreases. Maturity The probability that an individual matures p m (L) is deter- mined by size. Once an individual matures, she remains female until sex change (see below). We let L m represent the length at which 50% of the individuals will have matured. EiL)=aL h , (7) P,JL)- 1 where a and b are constants. Once an individual has changed sex (as determined by the sex change rule described above) sperm production (in millions) S(L) is given by S{L)=cL d , (8) l + exp(-q(L- L m (5) where c and d are constants. Size-dependent fecundity has been measured in many fish species (e.g. Gunderson, 1997). A general allometric relationship between sperm production and size has not been established. Therefore, we assumed that male gamete production increases with size at the same rate as that for females ib=d). We also assumed that males produce many more sperm at any body length than females produce eggs. Clearly, other possible patterns exist. We examined the case where males produce from 10 2 to 10 6 sperm for every egg produced by a female. In the pelagic spawning wrasse (Thalassoma bifasciatum ), large males release ap- proximately 1000 times more sperm than females release eggs (Schultz and Warner, 1991; Warner et al., 1995). We used recently published data on sperm production and fertilization rates in the bluehead wrasse (Thalas- soma bifasciatum) to generate a biologically appropriate fertilization function for our model (Warner et al., 1995; Petersen et al., 2001). It is critical to consider a biologically appropriate form for the function to express fertilization rates when considering the potential for sperm limitation. The probability an egg will be fertilized is an increasing function of the number of sperm available for that mat- ing (Fig. 2). The number of eggs released per mating also affects the fertilization rate (Fig. 2). For simplicity, we cal- culated the average expected fertilization rate per mating site based on the total production of sperm and eggs at the site. We let S represent the number of sperm released (in millions) and £ the number of eggs released at each mating site. We assumed that the proportion of eggs fertilized per mating site p F is given by where q determines the steepness of the probability function. Sex change The probability of sex change, p c iL), is a logistic function of absolute size L P,.(L) = l + exp(-p(L-L, )) (6) where L r represents the size at which 50% of the indi- viduals will change sex from female to male and p is a constant. Reproduction We assumed that female fecundity E(L) depends on indi- vidual size according to the allometric relationship Pf l + iisE + X )S (9) where k and % are constants fitted to the data. The number of eggs fertilized per group is p h -E and the total production of fertilized eggs. Pit), is the sum of the number of eggs fertilized in all mating groups. Measures of spawning stock biomass per recruit To measure the impact of fishing on stock dynamics, we computed the total spawning stock biomass per recruit starting from the beginning of fishing for the next 50 years. We used the generally recognized pattern that fish wet weight tends to be approximately proportional to the cube offish length (Gunderson, 1997) to convert fish length, L, into relative biomass, B(L)~L\ Then we calculated total female and male spawning stock biomass Alonzo and Mangel: The effects of size-selective fisheries on the stock dynamics of and sperm limitation in sex-changing fish per recruit (SSBR). We also kept track of the total fecundity (egg production per recruit I, fertility (sperm production per recruit), and eggs fertilized per recruit. Marine reserves ^=150 (about 60 females) OS- 'S 0.6 ■s 0.4 02 We examined the effect of no-take marine reserves on the predicted stock dynamics by comparing the stock dynamics in the presence and absence of reserves. Without a reserve, individuals at all mating sites are subject to fishing. In the presence of a no-take marine reserve, we "protect" a percentage of the mating sites (and thus the population) from fishing. We examined cases in which 09c, 10%, 20%, and 30% of mating sites were protected from fishing. We assumed that the population is completely open among mating sites. Thus, eggs produced from all mating sites enter one larval pool and recruitment occurs randomly between mating sites. Clearly other possibili- ties exist and could be considered in future analyses, but this case represents a reasonable baseline situation to con- sider because many marine fish have pelagic larval phases. We also recognize that these analyses ignore the effect of interactions between species within the reserve on stock dynamics. We examined two situations. In the first case, reduced fishing effort occurs when mean fishing mortality is decreased in the presence of reserves because fishing mortality (F) at the unprotected sites remains the same as before the reserve. In the second case, the redistribution of fishing effort occurs when mean fishing mortality across all sites remains the same because fishing mortality increases at the unprotected sites. Comparison of sex-changing stocks and dioecious stocks Ideally, we would like to distinguish the effects of sex change in isolation from the confounding effects of mating pattern, sex ratio, survival, growth, and population fecundity on stock dynamics. To differentiate whether sex change in iso- lation or other aspects of the mating system determine the predicted stock dynamics, we also examined a version of the model described above for a population where sex is fixed at birth. In this dioecious population, we keep all aspects of the stock constant except for the pattern of sex determi- nation (whether the species changes sex or is dioecious). One would generally expect a dioecious population with no differences between the sexes in mortality to exhibit a 50:50 sex ratio ( Fisher, 1930; Trivers, 1972; Charnov, 1982 ). However, we wanted to control for all differences between the dioecious and protogynous stocks other than the sex- determination pattern. Therefore, we considered the same sex ratio at maturity (0.67=the proportion of adults that are female) as found in the sex-changing population in the absence of fishing. Assuming no sex-specific differences in survival to maturity, this is the same as assuming a 0.67 sex ratio at birth. In this model, individuals remain one sex (determined randomly at birth) throughout their lifetime. K m =1750 (about 700 females) K>750 (about 300 females) 5.000 10.000 1 5.000 20,000 Sperm number (S) (in millions, about 1 to 100 males) Figure 2 Fertilization rate as a function of the number of eggs and sperm per mating site. The saturation parameter K m =\E+x is taken from Equation 9. Fishing is size but not sex selective. We assumed that males mature at the same size as females. Parameter values We used previous research on California sheephead (Lab- ridae, Semicossyphus pulcher), a commercially important sex-changing fish, to provide evolutionarily and ecologi- cally reasonable parameters for the model. Although the growth, survival, and reproduction of this species have been studied, less is known about the factors that induce sex change and mating behavior. In this species, sex change occurs at approximately 30 cm although the exact pattern varies among populations (Warner, 1975; Cowen, 1990). It is not known whether sex change is fixed or socially medi- ated. Because nothing is known about fertilization rates in the California sheephead, we generated k and y L c ), the general patterns remain the same, but for the same fishing mortality (.F), the effect of fishing on the population is less (Fig. 4). Female biomass does not decrease much with fish- ing mortality when L f =L c even though some females are removed by the fishery because the probability of a female changing sex is the probability of it being fished. Therefore, female loss due to the fishery affects male biomass rather than female biomass in the population. Sperm limitation and production The removal of large males from the population is pre- dicted to cause sperm limitation and decreased fertiliza- tion rates (Fig. 3, A and C), leading to a decrease in mean population size (Fig. 4A). The degree to which the fertiliza- tion rate and thus the population size decreases depends to a great extent on the pattern of sperm production and fertilization. We assumed that only a few males are needed to fertilize the eggs of many females (Fig. 2). We also assumed that per-capita reproduction and recruitment are high even at a low population size (Barrowman and Myers, 2000). Thus, protogynous populations with lower sperm production or fertilization rates would experience greater effects from fishing than predicted in the present study. Similarly, populations with lower production or sur- vival would experience larger decreases in population size even with the same level of sperm limitation and fishing. In general, however, the removal of males alone from a pro- togynous population with a fixed sex change is predicted to cause decreased fertilization rates and lower mean popula- tion size even when the fertilization rate function is asymp- totic and individual male sperm production is high. Mating aggregation size As mating aggregation size decreased and fishing mortality and effort remained constant, the effect of fishing on the pop- Eggs produced 1 1.5 2 Fishing mortality (F) Figure 3 Spawning-per-recruit measures. Results are presented for the sex-changing stock with one mating site when L^= L c and r=l. Means across 20 simulations are given. For details see the general text. ulation increased. As described above, we assumed that fish- ing effort would not be concentrated on the few large mating aggregations and thus increase total fishing mortality. The sex ratio, mean size, mean fecundity, and mean fertility all remained the same across different mating aggregation sizes with constant fishing mortality. However, the mean fertilization rate and number of fertilized eggs per recruit decreased with mating group size ( Fig. 5 ) even though male biomass and SSBR remained the same. Both predicted mean population size and biomass taken decreased as fish- ing mortality increased (Fig. 5). This pattern was generated by sperm limitation in small mating groups. Smaller groups have higher probabilities that sperm production within the group will not be sufficient to fertilize the eggs produced within the mating group. Small mating aggregations may not only be sperm limited but also be male limited and fail to reproduce completely; populations with small group sizes (50 individuals or less) were predicted to become extinct in Fishery Bulletin 102(1) 5-25% of the simulations as fishing mortality (F) increased from to 1. The impact of mating group size on stock dynam- ics is thus predicted to be nonlinear. A threshold mating aggregation size appeared to exist below which sperm limi- tation and reproductive failure become common. Spawning-per-recruit measures For size-selective fishing, the spawning stock biomass per recruit of females is not predicted to decrease significantly with increased fishing mortality as long as some male size classes escape fishing (Lr>L v ). However, male biomass per recruit and sperm production per recruit are both predicted to decrease. Although egg production is not predicted to 900 800 A L,>L C CD n 700 to ^\ L,=L C <= 600 J o 'ra 500 " 3 g- 400 " Q. c 300 " ra | 200 ' \l,, ra £? 300.000 D T3 a a> c > < ° 200.000 CD 100,000 U/< 0.5 1 15 2 2.5 3 Fishing mortality (F) Figure 4 The effect of size-s ilect ive fishing on stock dynamics. We present results for the sex-changing stock with one mating site when r=l. Means across 20 simulations are given. For details see the general text. decrease with increasing size-selective fishing pressure, the number of fertilized eggs is predicted to decrease. When all male size classes are fished iL.>L c ), the stock is predicted to crash and therefore clearly female biomass and egg production are predicted to decrease with fishing mortality. In general, the predicted decrease in mean popu- lation size and reproduction is driven for the most part by decreased sperm production and consequently a reduction in the number of eggs fertilized per recruit. The relation- ships between fishing pressure and the classic spawning- per-recruit measures do not indicate the true effect that fishing is predicted to have on the protogynous population (Fig. 6). When L f >L c , female spawning stock biomass per recruit and eggs produced per recruit showed almost no effect of fishing on the population, even as mean population size decreased. Because of the size-selec- tive fishing pattern, total and male biomass per recruit decreased with fishing mortality and decreasing mean population size. However, male and total biomass per recruit did not reflect the increased effect of fishing on populations with smaller mating aggregations. The production of fertilized eggs per recruit decreased with increased fishing pressure and decreased more sharply for smaller mating aggregations. Only the number of fertilized eggs per recruit could assess the predicted effect of fishing on the protogynous population. Thus, classic SPR measures were predicted to fail in the presence of sperm limitation to assess the impact of fishing on a protogynous stock. Marine reserves and fishery management In the situation considered in this study, the pattern of fishing is more important to stock dynamics than the presence of marine reserves. We assumed a size- selectivity that allowed on average 50% of individuals of sex-changing size to escape the fishing gear. Thus, although the sex ratio does increase (become more female) by 20-40%, all males are not lost from the population (when L f s.L t . and r=l ). If fishing selectivity occurs at a smaller size, then the effects on the popula- tion are predicted to be much greater and the protogy- nous stock would suddenly become more affected than the dioecious population. For example, at L^=25 cm the protogynous stock is predicted to crash whenever F^l. This occurs not because of a reduction in the produc- tion of eggs but rather because of a failure to fertilize the eggs produced by surviving females. When males of all size classes are fished, populations can become male limited and fertilization rates drop drastically. A decrease in the production of fertilized eggs can lead to a decrease in female biomass, but it is the removal of males rather than females that causes this decline. When fishing effort is not redistributed after the formation of a reserve, the impact of fishing on the mean population size and SPR measures is predicted to decrease (e.g. Fig. 7A). However, if fishing effort is redistributed among unprotected areas, the benefit of the reserves to the protogynous stock decreases (Fig. 8A). Protecting some sites allows large males to Alonzo and Mangel: The effects of size-selective fisheries on the stock dynamics of and sperm limitation in sex-changing fish 9 escape fishing and thus increases the pro- duction of fertilized eggs at the population level. However, yield decreased proportion- ally to the percentage of sites protected by the reserve unless fishing effort is redis- tributed among the remaining sites. We as- sumed that fish do not move between sites after the larval stage, and thus larger and older individuals do not leave the reserve and become exposed to fishing. Although this assumption is clearly appropriate for some species, it is important to realize that the dy- namics and predictions would differ for more closed populations or migratory species. For the fishing pattern and biological scenario examined in this study, marine reserves are not predicted to increase biomass available to the fishery (Figs. 7B and 8B). Dynamics of dioecious versus protogynous stocks In the dioecious stock with a single ran- domly mating aggregation, both male and female biomass per recruit and fecundity or fertility per recruit are predicted to decrease as fishing mortality increases ( Fig. 6). Because both egg production and sperm production decrease with increased fishing pressure in the dioecious stock, the number of eggs fertilized per recruit did not differ much from the other SPR measures. Thus, SSBR and eggs per recruit also indicated the impact of fishing on the stock in dioecious stocks with large mating aggregations. The percent drop in population size and fertil- ized egg production is predicted to be much greater in dioecious species and occurred more quickly than in the sex-changing stock because of a reduction in overall population fecundity even in the absence of decreased fertilization rates. However, dioe- cious stocks are predicted to exhibit larger mean population size for the same fishing mortality and to support a larger fishery because of the additional egg production of large fecund females. At very small mating aggregations, sperm limitation is predicted even in the dioecious stock and fertilized eggs per recruit become a better indicator of stock dynamics in the presence of fishing. Dioecious stocks are also predicted to benefit from marine no-take reserves through the protection of large fecund females ( Fig. 7 ). Discussion In this study we developed a general frame- work that examines the consequences to 0.95 0.9 0.85 Egg production (per recruit) Fertilized eggs (per recruit) Mean population size Figure 5 Mating aggregation size affects the response to fishing. Large (one large mating aggregation ) and small ( 10 smaller mating aggregations I situations are compared. Percent change in the presence of fishing (from F=0 to F=l> in egg production per recruit, mean fertilized egg production per recruit, and mean population size are given. Total population fecundity and mean body size are lower for the smaller mating aggregations. PROTOGYNOUS POPULATION 1 1 F=3 Eggs produced F=0 r& S 0.9- 0.8- Eggs fertilized _n .a St' »■" 0.7- a® x o Eggs produced and fertilized DIOECIOUS POPULATION 0.6- 0.5. ft ti * 0.4 F=3 F=0 600 650 700 750 800 850 900 950 Mean population size Figure 6 Spawning-per-recruit (SPR) measures in a protogynous (squares) and dioe- cious (triangles) stock: Mean egg production per recruit (filled) and mean fertilized eggs per recruit (open) are shown for a randomly mating popula- tion with one large mating group. Error bars indicate the standard error of the mean. For the dioecious population, the two SPR measures overlap. 10 Fishery Bulletin 102(1) fisheries management of a behaviorally and evolution- ary reasonable life-history and sex-change pattern. We based our assumptions and parameter values on patterns observed in natural populations that have presumably evolved given the life history tradeoffs and expected repro- ductive success associated with these behaviors. However, we made various assumptions that affect the predicted patterns such as a fixed sex-change pattern, male mating success proportional to sperm production, and a very resil- ient recruitment function. Despite these assumptions, a number of general patterns emerge. Life-history pattern is important but not sufficient to predict stock dynamics In general, we predicted that a protogynous stock with fixed sex change will respond to the same fishing pressure o o ^ fl Q.— ' °> S3, -= cr> X 0) (yi + 7 2 s, <0) and t, = s, - 7i > 0) (s, -/, <0) (7) (8) where t i = dRIRS on the /th day from the first sampling; Yv Yz = coefficients of the equations; and Sj = dRIRL on the /th day from the first sampling. Model estimation Likelihood function The location and scale parameters at the first sampling (o and fe ), the coefficients of Equa- tion 6 (s max , a, and p k ), and the coefficients of Equations 7 and 8 (y-j and y 2 ) are estimated as values that maximize total log-likelihood. The total log-likelihood is evaluated by the adequate probability density function selected in the first step. The log-likelihood functions take the follow- ing forms: Normal distribution log, L„ ormal (a Q ,b„, s max , a j , p k , y v y 2 ) = X2>g* -A r exp[-(Z <7i -a,)/24 2 ] 2nb (9) Largest extreme value distribution s, =s max / 1 + exp 2>a+£a b * (61 where s i = dRIRL on the /th day from the first sampling; s max = potential maximum dRIRL of the animal; a., P k = coefficients of each independent variable; A = categorical variable ( a dummy variable indi- cating animal ages ) that takes the value 1 orO; E kl = the kt\\ environmental factor on the /th day from the first sampling; n A = number of age categories; and n E = number of environmental factors. The categorical variable takes the value of 1 when the animal is the category, otherwise it takes 0. The multivari- ate logistic function with s max = 1 is used for logistic regres- sions (Sokal and Rohlf, 1995). A method of giving a value to the categorical variable is described by Zar ( 1999). Modeling the change in scale The daily relative increase rate of scale parameter (dRIRS) and dRIRL must be cor- log e -L,ar gcs /o ,fe ,s max ,a,,^„7 1 ,)' 2 ) N n q =XZ 1 °g«{ (1/ V ex p[-^-«,> / 4] xexp{-exp[-(Z 9i -<5 9 )/feJU, (10) where a , 6 = values of the location and scale param- eters, respectively, at the first sampling; s max> a j> Pk = coefficients of Equation 6; Y v y 2 - coefficients of Equations 7 and 8; N = number of samplings; n q = number of data at the qth sampling; a q = location parameter at the qth sampling estimated by Equation 5 (r,=s, ); b q = scale parameter at the qth sampling esti- mated by Equation 5 (r~^); and / = length of the /th individual at the fiO ; M 4^%, -^— Turbidity -•— Salinity -■ 40 - 1/ \ 20 - l"'"l I 1 1 1 ~wj Mode (estimated by model 4.1) 90% confidence interval (estimated by model 4.1) ° Mode (sample) Date Figure 3 Environmental fluctuations and prediction of the growth oiCorbiculajapon- ica juveniles spawned in 1997 in Lake Abashiri by the best model (Model 4.1 in Tablel). (Al Insignificant environmental factors (factors excluded in the model selection), turbidity (equivalent to kaolin density, ppmi and salinity (psu, practical salinity unitl. (Bl Significant environmental factors (factors included in the model selection I, temperature (°C) and water fluorescence (equivalent to uranin density, /'g/L>. (Cl Daily relative increase rate of loca- tion parameter (dRIRLl and daily relative increase rate of scale parameter (dRIRS) estimated by the model. (Di Growth of Corbicula japonica; verti- cal bars represent 90% confidence intervals for the shell lengths of the samples. length distribution becomes asymmetric during growth, skcwness of the distribution would increase according to growth. However, there is no correlation between the skewness and the means of the shell lengths. Therefore, we thought that the shell length distribution of the cohort was already asymmetric just after settlement. Such a distribu- tion might be influenced by fluctuations in larval settle- ment during the spawning season; and larval settlement would be influenced by fluctuations in larval supply from the water column. During the spawning season of 1997. the average planktonic larval density gradually increased from 26 ind/m 3 on 25 July to a maximum of 603 ind/m 3 on Baba et al.: An environmentally based growth model for |uvenile Corbicula japonica 21 Table 2 95% confidence limits of location and scale parameters at the first sampling and coefficients of the best model constructed based on the largest extreme value distribution (models 4.1 in Table 1) estimated by profile likelihood method. dRIRL = daily relative increase rate of location parameter. dRIRS = daily relative increase rate of scale parameter. Temp. = water temperature, WF = water fluorescence, Sal. = salinity, Turb. = turbidity. Parameters at 1st sampling Max. dRIRL Age categorization Environmental factors Expressing relationship between dRIRS and dRIRL A, a, a. Temp. ft WF ft Sal. ft Turb. ft Lower 95 % Upper 95 % 0.294 0.304 0.039 0.045 0.010 -26.6' 0.013 -11.5' -14.6 -6.4 0.41 1.00 0.27 0.64 0.0027 0.0039 0.734 0.793 1 One common coefficient for the two categorical variables. 13 August. Then it sharply decreased to 3 ind/m 3 on 19 August (Baba et al., 1999). Such a pattern of larval-density fluctuation might have caused the asymmetric distribution of shell lengths of the settled juveniles. Another possible factor that influenced the shapes of the shell length distri- butions and the relationship between dRIRL and dRIRS is size-dependent mortality, e.g. predations and fisheries. Size-dependent mortality has been reported in several marine bivalves (e.g. Nakaoka, 1996). Potential predators of C.japonica are fishes, such as Japanese dace (Tribolo- don hakonensis) (also known as big-scaled Pacific redfin, FAO), Pacific redfin (Tribolodon brandtii), common carp (Cyprinus carpio), and the So-iny mullet (Liza haemato- cheila ) (Kawasaki 4 ). In our study, the size-dependent mor- tality was negligible because the range of the shell lengths observed in this study was very narrow. The shape of the distribution to describe a single cohort should be determined from the data. In contrast, single cohorts are usually separated from multicohort data by as- suming a normal distribution of lengths in a single cohort (e.g. Fournier and Sibert, 1990). Therefore, it is possible that multicohort analysis done without selection of an adequate distribution to describe a single cohort causes substantial bias in estimations of various stock features of animal populations, such as age composition, growth, mortality, and recruitment. In our preliminary analyses, we also tested smallest extreme value distribution, inverse Gaussian distribution, and lognormal distribution. The in- verse Gaussian distribution was the best for two samples; the lognormal distribution, was the best for two samples; the largest extreme value distribution was the best for ten samples. Therefore, it is reasonable to select the largest ex- treme value distribution. We selected a single distribution for our analyses, otherwise a discontinuous point would have appeared in the growth curve. Relatively large confidence intervals were obtained in the coefficients of the linear component of Equation 6, i.e. a , and /3 ; , (Table 2). The relatively large confidence inter- vals may indicate that the number of estimated coefficients is somewhat larger than the number of samplings. There- fore, to estimate these coefficients more precisely, we may need to investigate more cohorts spawned in other years in future investigations. Growth of C. japonica We identified extremely slow growth in C. japonica juve- niles, which grew to a modal shell length of 0.7 mm during the first year in Lake Abashiri, which lies at 43.7°N. Spats of C. japonica collected from 1992 to 1997 in Lake Shinji, which lies at 35.5°N, grew to a mean shell length of 6.7 mm in natural conditions by the first winter (Yamane et al. 2 ). Using environmental factors measured in Lake Shinji from 1990 to 1998 at monthly intervals (Seike 5 ), we simu- lated the growth of C. japonica with model 4.1. Corbicula japonica grew to a mean shell length of 1.4 mm (standard error, 0.37 ) by the first winter in the simulations. Therefore, the large difference in juvenile growth between the two habitats cannot be explained by environmental differences because the results of the simulation were apparently an underestimate. We think that the extremely slow growth of the juveniles (prolonged phase of meiobenthic develop- ment ) in Lake Abashiri is probably a geographical varia- tion, which is genetically determined, within C. japonica. However, there remains a possibility that the juvenile growth differences depend on other environmental factors not measured in this study. Therefore, the geographical 4 Kawasaki, K. 1997. Lagoon structure and fish produc- tion in Ogawara-ko Lagoon. /;; Final reports on fisheries in Ogawara-ko Lagoon (Tohoku Construction Corporation ed.), p. 4-33. Unpubl. rep. Construction Office for Takasegawa General Development of Tohoku Regional Construction Bureau, 3 Ishido, Hachinohe, Aomori 039-1165, Japan. 6 Seike, Y. 1990-98. Gobiusu: monthly report of water quality in Lake Shinji and Lake Nakaumi. Unpubl. rep. Faculty of Science and Engineering. Shimane University, 1060 Nishi- kawatsu, Matsue, Shimane 690-0S23, Japan. 22 Fishery Bulletin 102(1) 10 Sep 1997; mode: 0.30mm, scale: 0.04, n=341 13 May 1998; mode: 0.41mm, scale: 0.06, n=38 11 Jun 1998; mode: 0.51 mm, scale: 0.12, n=292 10 Jul 1998; mode: 0.57mm. scale: 0.10, n=610 13 Aug 1998; mode: 0.64mm, scale: 0.12. n=456 1 1 Sep 1998; mode: 0.70mm. scale: 0.17, n=202 14 Oct 1998; mode: 0.76mm. scale: 0.17. n=162 0.0 22 Apr 1999; mode: 0.74mm. scale: 0.15, n=265 + + 13 May 1999; mode: 0.81mm, scale: 0.20. n=241 0.2 t t H 0.1 00 28 Jul 1999; mode: 2.14mm, scale: 1.06, n=63 ^#T>fffi^^ 3456 0123456 Shell length (mm) Figure 4 Shell-length compositions of a single cohort of Corbicula japonica spawned in 1997. The raw data (shell lengths) are shown by +. The largest extreme value distribution estimated by the best model ( model 4. 1 in Table 1 1 is shown by a solid line. The largest extreme value distribution independently fitted by the maximum likelihood method is shown by a dashed line. The sampling date and values of location parameter I mode) and scale parameter independently fitted by the maximum likelihood method are shown in each panel. variation should be validated by reciprocal transplanta- tions or laboratory experiments (or both) in future inves- tigations. Prolonged phases of meiobenthic development have been reported in some marine bivalves (Nakaoka, 1992; Harvey and Gage, 1995). However, a prolonged phase of meiobenthic development as a geographical variation is rarely reported. In many species of bivalve, populations from higher lati- tudes have a slower initial growth rate; but longevity and ul- timate size in these populations are frequently greater than at lower latitudes (Newell, 1964; Seed, 1980). The extremely slow growth of C. japonica juveniles in Lake Abashiri may be an extreme example of this phenomenon. In Lake Abashiri, C. japonica failed to spawn in ten out of 21 years for which Baba et al An environmentally based growth model for juvenile Corbicu/a japonica 23 data were available because of low water temperatures dur- ing the summer spawning season (Baba et al., 1999). This means that a long life span is essential to sustain popula- tions of C. japonica in northern habitats. We think that a long life span is the ultimate factor for the extremely slow growth rate of C. japonica juveniles in Lake Abashiri. The growth response of C. japonica juveniles is much less susceptible to environmental factors before the second win- ter than after and is the proximate factor for an extremely slow growth rate. Nuculoma tenuis, a detritus feeder, de- velops its palp proboscides, its feeding apparatus, during the prolonged phase of meiobenthic development (Harvey and Gage, 1995). The change of growth susceptibility to en- vironmental factors in young ages may suggest that some functional morphological changes occur in C. japonica, also a filter feeder. In our preliminary analyses, we could not find a better model when we used different values of s max in Equation 6 between ages instead of categorical variables indicating ages. Therefore, we conclude that the difference in growth rates between ages is not due to a difference in potential maximum growth rate, at least in the range of the shell length observed in our study. When our model is ap- plied to a wider range of the shell lengths or other species, it is best to examine the age dependence of s max . Acknowledgments We express our thanks to T Kato, Vice-Head of the River Improvement Section in the Abashiri Local Office of the Hokkaido Development Bureau, for providing environmen- tal data on Lake Abashiri. We also thank the reviewers of Fishery Bulletin for providing helpful suggestions on our manuscript. Literature cited Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. In Second international symposium on information theory (B. N. Petrov and F. Csaki, eds. ), p. 267-281. Akademiai Kiado, Budapest. Baba. K.. T. Kawajiri. and Y. Kuwahara. 1999. Effects of water temperature and salinity on spawning of the brackish water bivalve Corbicula japonica in Lake Abashiri, Hokkaido, Japan. Mar. Ecol. Prog. Ser. 180: 213-221. Bayne, B. L. 1993. Feeding physiology of bivalves: time dependence and compensation for changes in food availability. In Bivalve filter feeders and marine ecosystem processes (R. F. Dame ed. ). p. 1-24. Springer Verlag. New York, NY. 1998. The physiology of suspension feeding by bivalve molluscs: an introduction to the Plymouth "TROPHEE" workshop. J. Exp. Mar. Biol. Ecol. 219:1-19. Bayne, B. L., and R. C. Newell. 1983. Physiological energetics of marine molluscs. In The Mollusca, 4(1) (A. S. M. Saleuddin and K. M. Wilbur eds.), p. 407-515. Academic Press, New York, NY. Burnham. K. P., and D. R. Anderson. 1998. Model selection and inference, 349 p. Springer, New York, NY. Campbell. D. E., and C. R. Newell. 1998. MUSMOD, a production model for bottom culture of the blue mussel, Mytilus edulis L. J. Exp. Mar. Biol. Ecol. 219:171-203. Dame, R. F. 1993. The role of bivalve filter feeder material fluxes in estua- rine ecosystems. In Bivalve filter feeders in estuarine and coastal ecosystem processes: NATO ASI Series (R. F. Dame ed. ). p. 245-269. Springer- Verlag, Berlin, Heidelberg. Evans, M., N. Hastings, and B. Peacock. 1993. Statistical distribution, 170 p. John Wiley & Sons, New York, NY. Fournier, D. A., and J. R. Sibert. 1990. MULTIFAN a likelihood-based method for estimat- ing growth parameters and age composition from multiple length frequency data sets illustrated using data for south- ern bluefin tuna iThunnus maccoyii). Can. J. Fish. Aquat. Sci. 47:301-317. Grant, J., and C. Bacher. 1998. Comparative models of mussel bioenergetics and their validation at field culture sites. J. Exp. Mar. Biol. Ecol. 219:21-44. Grant, J., M. Dowd, K. Thompson, C. Emerson, and A. Hatcher. 1993. Perspectives on field studies and related biological models of bivalve growth and carrying capacity. In Bivalve filter feeders in estuarine and coastal ecosystem processes: NATO ASI Series (R. F. Dame ed. I, p. 371-420. Springer- Verlag, Berlin, Heidelberg. Harvey, R., and J. D. Gage. 1995. Reproduction and recruitment of Nuculoma Tenuis (Bivalvia: Nuculoida* from Loch Etive. Scotland. J. Moll. Stud. 61:409-419. Heral. M. 1993. Why carrying capacity models are useful tools for management of bivalve molluscs culture. In Bivalve filter feeders in estuarine and coastal ecosystem processes: NATO ASI Series (R. F. Dame ed. ), p. 455-477. Springer- Verlag, Berlin, Heidelberg. Jorgensen, C. B. 1996. Bivalve filter feeding revisited. Mar. Ecol. Prog. Ser. 142:287-302. Kafanov. A. I. 1991. Bivalves on continental shelf and continental slope of Northern Pacific, p. 81-82. Science Acad. USSR, Vladivostok. MacMahon, R. F. 198.3. Ecology of an invasive pest bivalve, Corbicula. In The Mollusca, 61 12) (W. D. Russell-Hunter, ed.), p. 505-561. Academic Press, Orlando, FL. Nakamura, M., M. Yamamuro, M. Ishikawa, and H. Nishimura. 1988. Role of the bivalve Corbicula japonica in the nitrogen cycle in a mesohaline lagoon. Mar. Biol. 99: 369-374. Nakaoka, M. 1992. Age determination and growth analysis based on external shell rings of the protobranch bivalve Yoldia notabilis Yokoyama in Otsuchi Bay. Northeastern Japan. Benthos Res. 43:53-66. 1996. Size-dependent survivorship of the bivalve Yoldia notabilis (Yokoyama, 1920): the effect of crab predation. J. Shellfish Res. 15:355-362. Newell, G. E. 1964. Physiological aspects of the ecology of intertidal molluscs. In Physiology of Mollusca (K. M. Wilbur and C. M. Yonge eds. ), p. 59-81. Academic Press, London. Pauly, D. 1987. A review of the ELEFAN system for analysis of length 24 Fishery Bulletin 102(1) frequency data in fish and aquatic invertebrates. In Length-based methods in fisheries research (D. Pauly and G. R. Morgan eds.), p. 7-34. ICLARM conference pro- ceedings 13. 688 International Center for Living Aquatic Resource Management. Manila, Philippines and Kuwait Institute for Scientific Research. Safet. Kuwait. Sakamoto, Y, M. Ishiguro, and G. Kitagawa. 1983. Statistics based on information amount (Jouhour- you Toukei Gaku), 236 p. Kyoritu Shuppan, Tokyo. [In Japanese.] Scholten, H., and A. C. Smaal. 1998. Responses of Mytilus edulis L. to varying food concen- trations: testing EMMY, an ecophysiological model. J. Exp. Mar. Biol. Ecol. 219:217-39. Seed. R. 1980. Shell growth and form in the Bivalvia. //; Skeletal growth of aquatic organisms (D. G. Rhoads ed.), p. 23-67. Plenum Press, New York, NY. Sellmer, G. P. 1956. A method for the separation of small bivalve molluscs from sediments. Ecology 37:206. Sokal, R. R. and F. J. Rohlf. 1995. Biometry, 3rd ed., 887 p. W. H. Freeman and Com- pany, New York, NY. Utoh, H. 1981. Growth of the brackish water bivalve, Corbicula japomca Prime, in Lake Abashiri. Sci. Rep. Hokkaido Fish. Exp. Stn. 23:65-81. [In Japanese.] Yamakawa, T.. and Y. Matsumiya. 1997. Simultaneous analysis of multiple length frequency data sets when the growth rates fluctuate between years. Fish. Sci. 63:708-714. Yamamuro, M., and I. Koike. 1993. Nitrogen metabolism of the filter-feeding bivalve Corbicula japonica and its significance in primary produc- tion of a brackish lake in Japan. Limnol. Oceanogr. 38: 997-1007. Zar, J. H. 1999. Biostatistical analysis. 663 p. Prentice Hall, Upper Saddle River, NJ. 25 Abstract— Information is summarized on juvenile salmonid distribution, size, condition, growth, stock origin, and species and environmental associations from June and August 2000 GLOBEC cruises with particular emphasis on differences related to the regions north and south of Cape Blanco off Southern Oregon. Juvenile salmon were more abundant during the August cruise as compared to the June cruise and were mainly distributed northward from Cape Blanco. There were distinct differ- ences in distribution patterns between salmon species: chinook salmon were found close inshore in cooler water all along the coast and coho salmon were rarely found south of Cape Blanco. Dis- tance offshore and temperature were the dominant explanatory variables related to coho and chinook salmon distribution. The nekton assemblages differed significantly between cruises. The June cruise was dominated by juve- nile rockfishes, rex sole, and sablefish, which were almost completely absent in August. The forage fish community during June comprised Pacific herring and whitebait smelt north of Cape Blanco and surf smelt south of Cape Blanco. The fish community in August was dominated by Pacific sardines and highly migratory pelagic species. Esti- mated growth rates of juvenile coho salmon were higher in the GLOBEC study area than in areas farther north. An unusually high percentage of coho salmon in the study area were preco- cious males. Significant differences in growth and condition of juvenile coho salmon indicated different oceano- graphic environments north and south of Cape Blanco. The condition index was higher in juvenile coho salmon to the north but no significant differences were found for yearling chinook salmon. Genetic mixed stock analysis indicated that during June, most of the chinook salmon in our sample originated from rivers along the central coast of Oregon. In August, chinook salmon sampled south of Cape Blanco were largely from southern Oregon and northern Cali- fornia; whereas most chinook salmon north of Cape Blanco were from the Central Valley in California. Manuscript approved for publication 30 June 2003 by Scientific Editor. Manuscript received 20 October 2003 at NMFS Scientific Publications Office. Fish. Bull 102:25-46 (2004). Juvenile salmonid distribution, growth, condition, origin, and environmental and species associations in the Northern California Current* Rick D. Brodeur Northwest Fisheries Science Center National Marine Fisheries Service, NOAA 2030 S. Marine Science Drive Newport, Oregon 97365 E-mail address: Rick-Brodeuriffinoaa-gov Joseph P. Fisher College of Ocean and Atmospheric Sciences Oregon State University Corvallis, Oregon 97331 David J. Teel Northwest Fisheries Science Center National Marine Fisheries Service, NOAA Seattle, Washington 98112 Robert L. Emmett Northwest Fisheries Science Center National Marine Fisheries Service, NOAA 2030 S Marine Science Drive Newport, Oregon 97365 Edmundo Casillas Northwest Fisheries Science Center National Marine Fisheries Service, NOAA Seattle, Washington 98112 Todd W. Miller Cooperative Institute for Marine Resources Studies Oregon State University Newport, Oregon 97365 The need to understand the direct and indirect linkages between oceano- graphic conditions and salmon sur- vival in the marine environment has increased with the listing of many West Coast salmon stocks as threat- ened or endangered. Recent studies have shown that long-term changes in climate affect oceanic structure and produce abrupt differences in salmon marine survival and returns (Francis and Hare, 1994: Mantua et al., 19971. A major regime shift in the subarctic and California Current ecosystems during the late 1970s may have been a factor in reducing ocean survival of salmon in the Pacific Northwest and in increas- ing marine survival in Alaska ( Hare et al., 1999). Fluctuations in mortality of salmon in the freshwater and marine environments have been shown to be almost equally significant sources of annual salmonid recruitment variability ( Bradford, 1995 ). Unlike in the freshwa- ter environment, the physical and bio- logical mechanisms and factors in the marine environment that cause mor- tality of salmon are largely unknown. Predation, inter- and intraspecific competition, food availability, smolt quality and health, and physical ocean conditions likely influence survival of salmon in the marine environment. Thus, increasing our understanding of nearshore ocean environments, their linkages to oceanographic conditions, and the role they play in salmonid survival, could provide management options for increasing adult returns. Characterization of the space-time vari- ability of the environmental conditions that smolts encounter when they enter the nearshore ocean, and the eventual survival of these smolts will allow us to identify which biotic and abiotic ocean conditions are correlated with various ocean survival levels. Many anadromous salmonid popula- tions along the west coast of the United States have declined over the last few decades (Nehlsen et al., 1991), and most stocks show a regional north-south pat- tern in degree of extinction risk (Kope and Wainwright, 1998). This pattern suggests that both marine habitat con- ditions and mesoscale climate patterns affect salmonid population status (e.g. Lawson, 1993). A dramatic example is the population trend of coho salmon (Oncorhynchus kisutch) along the Or- egon coast. Populations along the coast north of Cape Blanco (43°N) have exhib- ; Contribution number 364 of the U.S. GLOBEC program. NEP Office, Oregon State University, Corvallis. OR. 26 Fishery Bulletin 102(1) ited a strong decline in size and survival in the mid-1990s; whereas populations south of Cape Blanco have not shown this trend (Lewis 1 ). This finding suggests that these two populations have experienced different ocean conditions. The quality of the marine habitat (in terms of habitat complexity, prey density, and temperature) undoubt- edly influences fish growth and condition. Growth and indices of condition can be used as measures of habitat quality for juvenile salmon and to identify essential links between oceanographic conditions and survival of salmon populations during the critical juvenile life history phase. Measures such as growth (growth rate, size variation, and allometric relationships) (Lorenzen, 1996; McGurk, 1996) and accumulation of energetic reserves used in growth and sustenance during the low-productivity winter periods have been used previously to characterize habitat quality and to describe how it ultimately affects the individual and the population (Perry etal., 1996; Paul and Willette, 1997). Environmental factors are known to affect growth, repro- duction, survival, and ultimately population recruitment (Hinch et al., 1995; Marschall and Crowder, 1995; Fried- land and Haas, 1996). As such, fish condition, growth rate, and size in the pre-adult stages are parameters that can be used to identify the influence of natural and anthropogenic ocean conditions on marine survival. Much of our current knowledge of the dominant nekton of the pelagic ecosystem off the coasts of Oregon and Wash- ington is derived from a series of 17 cruises conducted by Oregon State University (OSU) from 1979 to 1985. These collections, consisting of >900 quantitative purse seine sets in the northern California Current, were made to examine geographic distributions and temporal trends of the domi- nant nekton and how these relate to physical and biotic conditions at the time of capture. The primary purpose of these cruises was to collect data for assessment of the abundance, distribution, growth, migration, and ecology of juvenile salmon in coastal waters. Data on the distribution, migration and growth of juvenile salmon from these cruises have been summarized in Fisher and Pearcy (1988; 1995). Pearcy and Fisher ( 1988, 1990), and Pearcy ( 1992). Analy- sis of the nonsalmonid data includes studies on their abun- dance and distribution (Brodeur and Pearcy, 1986; Emmett and Brodeur, 2000), feeding habits (Brodeur et al., 1987) and interannual variability in relation to oceanographic conditions (Brodeur and Pearcy, 1992). In addition, the distribution of juvenile salmon (mainly coho and chinook salmon [O. tshawytscha}) has been studied more recently as a component of a multiyear Columbia River Plume study (Emmett and Brodeur, 2000; Teel et al., 2003; Brodeur et al., 2003). However, all these cruises extended only as far south as Cape Blanco, with the exception of one cruise (July 1984), which extended as far south as Eureka, California, but included only a few collections south of Cape Blanco (Pearcy and Fisher, 1990). Thus, the region south of Cape Blanco is almost completely unknown in terms of juvenile 1 Lewis, M. A. 2002. Stock assessment of anadromous salmo- nids 2001. Monitoring program report OPSW-ODFW-2002-04, 57 p. Oregon Dept. Fish Wildlife, Portland. OR 97207. salmon distribution, pelagic nekton, and biological ocean- ography in general, despite being an area of very strong upwelling and high productivity. Also, the fine-scale dis- tribution of juvenile salmon in relation to environmental variables has not been studied in any detail. The California Current is not homogeneous but rather can be divided into distinct subunits or regions, each with its own physical and biological characteristics (U.S. GLO- BEC, 1994). A break between the northernmost two regions occurs at Cape Blanco, where the equatorward upwelling jet veers sharply off the shelf and into the California Cur- rent (Barth et al., 2000). The upwelling zone north of the cape is narrow, extending out about 30 km, whereas south of Cape Blanco, it can extend up to 100 km offshore. This area also appears to represent a faunal break for some zoo- plankton communities (McGowan et al., 1999; Peterson and Keister, 2002) and is a break point for alternative salmon migration strategies (Weitkamp et al., 1995; Weitkamp and Neely, 2002). During the summer of 2000, we conducted broad-scale sampling and fine-scale process studies from central Or- egon to northern California to examine the distribution of juvenile salmon and associated species in relation to environmental conditions. This was one component of a multidisciplinary U.S. Global Ocean Ecosystem Dynamics (GLOBEC) Northeast Pacific study examining the north- ern California Current ranging in scope from the physics up to the top trophic levels (Batchelder et al., 2002). We were interested in examining the distribution of juvenile salmon north and south of Cape Blanco, the origin of these fish, and any regional differences in growth and condition of salmon across the range of sampling. Evidence exists that the physical conditions and the associated biota are different within this geographical scale. Thus, analyses of the relationship between oceanographic conditions and the response of resident biota can provide insights into the linkages associated with physical and biological processes that shape the biological community, and in particular, those associated with salmon recruitment. Methods Field surveys Surveys were conducted over two time periods — early summer (29 May-18 June, 2000) and late summer (28 July-15 August, 2000). Each survey consisted of a meso- scale grid along designated GLOBEC transects that had been monitored for several years and by fine-scale pro- cess sampling at stations of interest based on features observed in the physical environment (fronts or eddies) or by acoustic sampling conducted by two accompanying oceanographic vessels (RV Wecoma and RV New Horizon). Further details on the physical and biological conditions occurring at the time of our sampling have been reported by Batchelder et al. (2002). For the mesoscale survey, stations were established at 1, 5, 10, 15, 20, 25 and 30 nautical miles from shore on each of five transects. Inclement weather, particularly Brodeur et al.: Distribution, growth, condition, origin, and associations of |uvenile salmonids 27 during the first cruise, prevented us from sampling all the stations along each transect. At each station, a Nordic 264 rope trawl built by Nor'Eastern Trawl Systems, Inc. (Bainbridge Island, WA) was towed in surface waters by a chartered fishing vessel (FV Sea Eagle) at a speed of 6 km/h. This rope trawl has a maximum mouth opening of approximately 30 m x 18 m. Mesh sizes ranged from 162.6 cm in the throat of the trawl near the jib lines to 8.9 cm in the codend. To maintain catches of small fish and squid, a 6.1-m long, 0.8-cm mesh knotless liner was sewn into the codend. All tows were 30 minutes in duration. All fish and squid caught were counted and measured at sea. After fork length (FL) was measured to the nearest mm, all juvenile salmon were immediately frozen for later determinations of growth, condition, food habits, genetic analysis, and as- sessment of pathological condition. The physical and biological environment was monitored and sampled at each station immediately prior to setting the trawl. A CTD (conductivity, temperature, and depth) cast was made with a Sea-Bird SBE 19 Seacat profiler to 100 m at deep stations or within 10 m of the bottom at shallow stations. Chlorophyll and nutrient samples were collected from 3 m depth with a Niskin water sampler. A neuston tow with a 1-m 2 mouth containing 333-,(im mesh net was towed for 5 minutes out of the wake of the vessel at each station. General Oceanics flow meters were placed inside the net to measure the amount of water sampled. Additional details on the analysis of these neuston trawls are available in Reese et al. 2 Condition and growth analysis Each salmonid was remeasured (FL to the nearest mm) and weighed (to the nearest 0.1 g) in the laboratory. A por- tion of hepatic and muscle tissue was excised, placed in individual capsules, frozen in liquid nitrogen, and stored at -80°C until analyzed. The bioenergetic health of juve- nile salmon was evaluated by assessing changes in water content (as a surrogate measure of fat accumulation) of liver and muscle to estimate dry tissue weight. The water content was determined by drying tissue samples to a con- stant weight at 105°C. The accumulation of energy reserves during the growth season ( energy reserves of salmon in August in relation to salmon collected in June) that would enhance survival of juveniles during the winter when food availability is lower was also measured. The condition of juvenile salmon was assessed by examining weight residu- als (by using either the wet weight or dry weight) derived from the allometric relationship between length and weight of individual juvenile salmon after logarithmic transforma- tion (Jakob et al., 1996) of salmon captured in June and August. Wet-weight residuals are representative of the traditional condition index of animals and are a reflection 2 Reese, D.C., T.W.Miller, and R.D. Brodeur. 2003. Community structure of neustonic zooplankton in the northern California Current in relation to oceanographic conditions. 22 p. Unpubl. manuscript. Northwest Fisheries Science Center, NMFS. 2030 S. Marine Science Drive, Newport, OR 97365. of somatic tissue growth. Dry-weight residuals are respon- sive to accumulation of fat stores and are a reflection of the bioenergetic health of the individual animal (Sutton et al., 2000; Post and Parkinson, 2001). To contrast growth characteristics during 2000 in differ- ent latitudinal ranges of the California Current, we com- pared ocean growth rates of juvenile coho salmon south and north of Cape Blanco in the GLOBEC study area, and in the area from Newport, Oregon, north to northern Washington. The physical and biological characteristics of these three regions of the coastal ocean differ greatly (U.S. GLOBEC, 1994), and these differences may impact the dis- tribution and abundance of prey of juvenile salmonids and therefore may also affect salmonid growth. Data north of Newport, Oregon, were collected during a separate study of the Columbia River plume and the adjacent coastal ocean (hereafter called the "plume study") using the same trawl and a similar sampling strategy as in the GLOBEC study (see Emmett and Brodeur [2000] and Teel et al. [2003] for details). Scales were examined from 45 juvenile coho salmon caught during the June and August 2000 GLOBEC cruises and 252 juvenile coho salmon caught during the 2000 plume cruises. The scales were mounted on gummed cards from which acetate impressions were made. Using a video camera attached to a compound microscope and Optimas® imaging software (vers. 5.1, Optimas Inc., Se- attle, WA) we measured the distance (scale radius) along the anterior-posterior axis of each scale from the focus (F) to the ocean entry mark (OE) and to the scale margin (Fig. 1). The fork-length of each fish at the time of ocean entry (FL 0E ) was estimated from the scale radius (SR 0E ) at ocean entry using the Fraser and Lee back-calculation method (Ricker, 1992): FL„ (FL- 36.07) SR xSR of . +36.07, where FL = length at capture; SR = scale radius at capture; and 36.07 = the intercept from a regression of SR on FL for juvenile coho salmon caught in the ocean (Fig. 2A). In an analogous fashion, fish weight at time of ocean entry (Wr 0£ ) was back-calculated f length at ocean entry (FL 0E ): (Wt 0E ) was back-calculated from the estimated fish fork \ni Wt 0E ) = (ln(Wr 1 + 12.633) ln(FL) xln(FL r , F 1-12.633, where Wt = weight at capture; and -12.633 = the intercept from a linear regression of ln(Wr) on ln(FL) for juvenile coho salmon caught in the ocean (Fig. 2B). The growth rate in FL, (FL-FL 0E )lAd, 28 Fishery Bulletin 102(1) Figure 1 Scale from a 352-mm FL male juvenile coho salmon (Oncorhynchus kisutch) caught during the August 2000 GLOBEC cruise showing the axis of measurement (black line), the focus (F), the mark of ocean entry (OE), and the scale margin (SM). and the instantaneous growth rate in weight: G = (MWt)-MWt 0E ))/M, where Ad = estimated days between ocean entry and cap- ture, were estimated for each salmon. The meaning of the instantaneous growth rate G can be stated as follows: if salmon growth is exponential between ocean entry and capture, then Wt Wt„ and at any instant the fish's weight increases at the rate of G of its body weight per day. G can be multiplied by 100 to give the instantaneous growth rate in terms of percentage of body weight per day. Although the dates of ocean entry of individual lish were unknown, seaward migration of coho salmon smolts in California, Oregon, and Washington rivers occurs mainly between mid-April and mid-June, and there is no consis- tent latitudinal trend in timing of the migration ( Weitkamp et al., 1995). Peak downstream migration of coho salmon smolts was between mid-May and very early June in the Columbia River estuary, 1978-83 (Dawley et al., 1985), and in the lower Trinity River, California, 1997-2000 (US- A FL (mm) vs scale radius (mm) GM Regression: FL = 152.22 SR +36.07 r 2 = 0.94, n=370 1 2 Scales radius (mm) B Wt(g)vs FL(mm) In(WI) = 3.2273'ln(FL) - 12 6329 or Wt(g) = 3.263x1 (T 5 FL(mm) 32273 n=1018V = 0.99 s — 5 In (FL) Figure 2 (A) Regression of fork length (FL) on scale radius and. 'Bi regression of ln(WY) on ln(FL) for juvenile coho salmon {On- corhynchus kisutch) caught during the May 1998-September 2000 Columbia River plume study. FWS 3 ). In 2000, peak downstream migration of mainly nonhatchery coho salmon smolts at 13 monitoring sites in coastal Oregon rivers north of Cape Blanco occurred from April 2 to May 20; median peak migration occurred 26 April ( Solazzi et al. 4 ) From the information available on timing of seaward migration of coho salmon smolts. we used an ocean entry date of 15 May when calculating Ad and estimating ocean growth rates of unmarked coho salmon from scales. In addition to estimating growth rates of juvenile coho salmon from scales, we also estimated instantaneous growth rates in weight between hatchery release and cap- ture in the ocean of 28 coded-wire-tagged (CWT) juvenile coho salmon: USFWS (U.S. Fish and Wildlife Service). 2001. Juvenile sal- monid monitoring on the mainstem Klamath River at Big Bar and mainstem Trinity River at Willow Creek, 1997-2000, 106 p. Annual report of the Klamath River Fisheries Assessment Pro- gram. Areata Fish and Wildlife Office, Areata, CA 9552 1 . Solazzi, M.F., S.L.Johnson, B.Miller, and T.Dalton. 2002. Sal- monid life-cvele monitoring project 2001. Monitoring program report OPSW-ODFW-2002-2, 25 p. Oregon Dept. Fish and Wildlife, Portland. OR 97207. Brodeur et al.: Distribution, growth, condition, origin, and associations of juvenile salmonids 29 G = (MWt)-MWt R ))/M, where Wt = weight of the CWT fish at capture; Wt R = the average weight of fish in the CWT group at time of release; and Ad = days between hatchery release and capture in the ocean. Estimated growth rates of these CWT fish, of known release date and known average release weight were used to vali- date the growth rates estimated from scale analysis Our analysis of the growth of chinook salmon based on scale characteristics is not far enough advanced to report in this article. We plan to present these data in a later article. Contribution of hatchery coho salmon to catches The total numbers, percentages of marked fish ( any exter- nal fin clips or internal tags) and grand average weights of hatchery coho salmon smolts released in 2000 are sum- marized for different release regions in Appendix Table 1. These data were used to compare the estimated average weights of fish at time of ocean entry (from scale analy- sis ) with the average weights of hatchery fish at time of release, and also to estimate the proportions of hatchery coho salmon in our catches. We calculated the expected percentage (E%) of marked fish in each catch if 100% of the fish were hatchery fish: E% X*.*4, where i?, = the proportional contribution of region i to the catch (this paper for the GLOBEC catches, and from Teel et al., 2003 for the plume study catches); and A, = the percentage of hatchery fish that were marked in region i ( from Appendix Table 1 ). The percentage of hatchery fish in each catch sample (H%) was then estimated as 0% H% = — xlOO, E% where O c A = observed percentage of marked fish. Genetic analysis The freshwater origins of juvenile chinook and coho salmon and steelhead (O. my kiss) were studied by using standard methods of genetic mixed stock analysis (Milner et al., 1985; Pella and Milner, 1987). According to the methods described by Aebersold et al. (1987), samples of eye, liver, heart, and skeletal muscle were extracted from frozen whole juvenile salmon and analyzed with horizontal starch-gel protein electrophoresis. Data from previous studies char- acterizing genetic (allozyme) differences among spawning populations in California and the Pacific Northwest were then used as baseline data to estimate the stock composi- tions of our ocean caught mixed-stock samples. Baselines consisted of 32 gene loci and 116 populations for chinook salmon (Teel et al. 5 ), 58 loci and 49 populations for coho salmon (Teel et al., 2003), and 55 loci and 57 populations for steelhead (Busby et al., 1996). Estimates of stock com- positions were made by using the maximum likelihood procedures described by Pella and Milner (1987) and the Statistical Package for Analyzing Mixtures (Debevec et al., 2000). Estimates of individual baseline populations were then summed to estimate contributions of regional stock groups. Precision of the stock composition estimates was estimated by bootstrapping the estimates 100 times with resampling of the baseline and mixture genetic data as described in Pella and Milner (1987). Habitat and assemblage analysis The raw numbers offish and squid caught from each trawl were converted to densities based on the volume filtered per trawl to standardize for differences in effort between tows. Density contours of juvenile salmon and other nekton were produced using specialized graphics programs. We then tested whether the habitat associations of the domi- nant salmonids were significantly different from the total habitat sampled by following the methods outlined in Perry and Smith ( 1994). This procedure involved comparing the cumulative distributions of salmon catch with observed environmental conditions (temperature, salinity, chloro- phyll-a at one meter, water depth, and neuston displace- ment volume). We performed 5000 randomizations of the data and used the Cramer-von Mises test statistic recom- mended by Syrjala ( 1996) as being robust to the effects of inordinately large catches. To explore the relationship between juvenile salmon and other fish species and environmental variables, we used several types of multivariate analyses (McCune and Grace, 2002 ). Original data from each of the two cruises formed complimentary species and environmental matrices. The June and August cruises were analyzed individually to look at spatial patterns of species composition in relation to environmental gradients (Gauch, 1982). To avoid spurious effects of rare species, we excluded species from the data matrix that had a frequency of occurrence of less than 10% of the possible occurrences for each cruise (McCune and Grace, 2002). To minimize the effect of very large catches, the data were log transformed. Stations with no species present were eliminated from the data set to allow for anal- ysis of sample units in species space. Data transformations and their effects on the summary statistics were examined prior to analysis. Analyses of data were performed by using PC-ORD version 4.28 (McCune and Mefford, 1999). Agglomerative hierarchical cluster analysis (AHCA) using the Bray-Curtis dissimilarity measure and Wards Teel, D. J„ P. A. Crane. C. M. Guthrie, III, A. R. Marshall. D. M. Van Doornik, W. D. Templin, N. V. Varnavskaya, and L. W. Seeb. 1999. Comprehensive allozyme database discriminates chinook salmon from around the Pacific Rim. (NPAFC docu- ment 440), 25 p. Alaska Department of Fish and Game, Divi- sion of Commercial Fisheries, 333 Raspberry Road, Ancorage, AK 99518. 30 Fishery Bulletin 102(1) linkage function was applied to arrange the nekton spe- cies assemblages and stations into cluster groups. The cutoff level to form optimal groups within the species and station dendrograms was based on several criteria: 1) biological meaning; 2) significance tests of groups using a multi-response permutation procedure (MRPP); and 3i comparison of cutoff level MRPP results with those groups obtained from one cutoff level below and above the level of interest. A nonparametric procedure, MRPP compares the a priori groupings from AHCA and tests the hypothesis of no difference between the groups. For cluster analysis of stations, indicator species analysis (ISA) was used to determine nekton species strongly associated with indi- vidual groups. ISA assigns indicator values to each spe- cies according to relative abundance and frequency, then tests the significance (Monte-Carlo permutation test) of the highest species-specific indicator value assigned to a particular group. Nonmetric multidimensional scaling (NMS; Kruskal, 1964) was used to ordinate sample units in species space and to compare station cluster groups to environmental gradients. NMS was chosen for this analysis because it is robust to data that are non-normal and that have high numbers of zeros. Initial runs of NMS from both cruise da- tasets resulted in three-dimensional solutions. Subsequent reapplication of NMS using a three-dimensional solution (Sorensen distance, 400 maximum iterations, and 40 runs with real data) was applied for the final ordinations. To examine the environmental or station factors associated with each NMS axis that may have affected the distribu- tion of the dominant taxa, we correlated the NMS station and species scores to a suite of environmental variables including water depth, distance offshore, latitude, surface temperature, surface salinity, chlorophyll-a concentration, and neuston zooplankton settled volumes. Pearson and Kendall correlations with each ordination axis were used to measure strength and direction of individual species and environmental parameters. Results Distribution of juvenile salmon and other species We collected a total of 18,852 nekton individuals: two ceph- alopod, one agnathan, two elasmobranch, and 57 fish taxa from 163 surface trawls (see Table 1 for scientific names of all species). With the exception of market squid in June and blue shark in August, most of the nonteleost nekton occurred in only a few collections. Substantially fewer fish were caught in the June cruise than in the August cruise, but the diversity was much higher in the June cruise. The catch in June was dominated by forage fishes such as Pacific herring, surf and whitebait smelt, and juvenile rock- fishes, sablefish, and flatfishes. Salmonids, mainly juvenile chinook and coho salmon and steelhead, comprised a rela- tively minor proportion of the catches (only 114 juvenile salmonids; 1.9 % of the total). The August cruise was dominated by several large catches of Pacific sardine (Table 1 ). Jack mackerel was the most common nonsalmonid caught. Many of the juvenile fish taxa caught during the June cruise were absent during the August cruise; those that did occur ( sablefish. rex sole) were much lower in abundance. Mesopelagic fishes of the family Bathylagidae and Myctophidae were collected only during the August cruise, mainly because of the inclusion of more offshore stations and occasional collections during nondaylight hours. As in the earlier cruise, salmonids com- prised a relatively minor percentage of the catch (3.19f ) but were more common and abundant during this survey. Juvenile chinook salmon were broadly distributed lati- tudinally during both cruises, but their distribution was mainly restricted to nearshore stations within the 100-m isobath (Fig. 3). Coho salmon juveniles were more common north of Cape Blanco during both cruises and were found generally farther offshore than chinook salmon juveniles (Fig. 3). In contrast, steelhead juveniles were found mainly south of Cape Blanco, especially in June, but their zonal distribution overlapped that of coho salmon juveniles. Size and condition of juvenile salmon Fork length of yearling chinook salmon averaged 227 ±42 mm FL in June and 230 ±30 mm FL in August and aver- aged 135 ±12 mm FL for subyearling chinook salmon in August, whereas juvenile coho salmon averaged 162 ±32 mm FL in June and 286 ±46 mm FL in August ( Table 2 ). No significant differences in fork length of juvenile chinook or coho salmon north or south of Cape Blanco were evident. Juvenile coho salmon weighed significantly more on a wet-weight basis for a given fork length in the region north of Cape Blanco compared to juveniles captured south of Cape Blanco (Fig. 4). This pattern was also similar and significant when evaluated on a dry-weight basis (bioen- ergetic growth). Although the stock composition in the two regions could account for some of these differences, the growth responses likely reflect habitat-specific features in the region north of Cape Blanco that benefit coho salmon. No difference in condition of yearling chinook salmon cap- tured north or south of Cape Blanco, on either a wet- or dry- weight basis, was evident (Fig. 4). Information regarding size and condition of subyearling chinook salmon are not presented because few subyearling chinook salmon were caught in June and all but one subyearling chinook salmon in August were caught in the region south of Cape Blanco, OR. Insufficient subyearling chinook salmon were avail- able for an analysis comparable to that done for yearling chinook and coho salmon. Proportions of wild and hatchery coho salmon Most of the juvenile coho salmon caught during the plume study north of Newport, Oregon, originated in hatcher- ies (Table 3). In June and September 2000 we estimated that wild fish comprised only W9i and 25 r < . respectively, of the catch. Wild fish, however, comprised a proportion- ally much higher percentage of the catch of coho salmon in the GLOBEC study area in June north of Cape Blanco (67$ I, and in August south of Cape Blanco (619! I, than in the plume study area farther to the north. Most jacks and Brodeur et al.: Distribution, growth, condition, origin, and associations of |uvenile salmonids 31 Table 1 Phylogenetic listing of nekton catch in numerical composition, frequency of occurrence (F.O.) and size range cau ght for each cruise. (j) indicates juvenile stage; (a) adult. ML = mantle length, TL = total length. FL = fork length, SL = standard length ( in mm). Class and Family Common name June (84 stations) August (79 stations) Scientific name dumber F.O. Size range Number F.O. Size range Cephalopoda Onychoteuthidae Pacific clubhook squid Onychoteuthis borealijaponicus 19 6 21-80 ML 302 6 21-227 ML Loliginidae Market squid Loligo opalescens 301 14 33-122 ML 1 1 35 ML Agnatha Petromyzontidae Pacific lamprey Lampetra tridentata 1 1 625 TL Chondrichthyes Alopiidae Thresher shark Alopias vulpinus 1 1 36-576 TL Carcharhinidae Blue shark Prionace glauca 18 10 1300-1660 TL Osteichthyes Xenocongridae Eel leptocephalus Thalassenchelys coheni 3 1 214-243 TL 2 2 260-305 TL Clupeidae Pacific herring Clupea pallasi 1022 9 127-195 FL Pacific sardine Sardinops sagax 7 2 237-260 FL 10,327 15 178-290 FL Engraulididae Northern anchovy Engraulis mordax 49 12 148-165 FL Salmonidae Chinook salmon (j,a) Oncorhynchus tshawytscha 56 18 121-780 FL 252 26 109-910 FL Coho salmon (j,a) Oncorhynchus kisutch 35 15 122-580 FL 111 25 210-736 FL Cutthroat trout (j,a) Oncorhynchus clarki 1 1 186 FL 3 3 258-341 FL Steelhead trout (j,a) Oncorhynchus mykiss 22 8 176-284 FL 36 13 261-430 FL Osmeridae Smelt (j) Osmeridae 14 4 37-52 SL 74 5 31-50 SL Surf smelt Hypomesus pretiosus 846 8 128-184 FL 351 7 140-187 FL Whitebait smelt Allosmerus elongatus 946 6 60-114 FL 79 3 76-132 FL Bathylagidae Popeye blacksmelt Bathylagus ochotensis 1 1 76 SL Paralepidae Slender barracudina Lestidium ringens 3 1 72-76 SL Myctophidae Northern lampfish Stenobrachius leucopsarus 96 4 14-70 SL Bigfin lanterfish Symbolophorus californiensis 61 4 89-102 SL Blue laternfish Tarletonbeama crenularis 10 3 33-87 SL Gadidae Gadid(j) Gadidae 10 3 42-58 SL 13 3 53-57 SL Pacific cod 1 j ) Gadus macrocephalus 23 1 38-60 SL Pacific tomcod ( j ) Microgadus proximus 6 4 35-55 SL 8 2 49-80 SL Scomberesocidae Pacific saury Cololabis saira 26 1 182-229 FL 66 6 131-194 FL Atherinidae Jacksmelt Atherinopsis californiensis 1 1 302 FL Trachipteridae King-of-the-salmon (j ) Trachipterus altivelis 2 2 71-270 SL 12 2 40-83 SL Gasterosteidae Threespine stickleback Gasterosteus aculeatus 1 1 60 SL Scorpaenidae Pacific ocean perch (j ) Sebastes alutus 1 1 33 SL Darkblotched rockfish (j Sebastes crameri 154 14 29-54 SL 1 1 53 SL Yellowtail rockfish (j) Sebastes flavidus 1350 24 20-63 SL 1 1 18 SL Shortbelly rockfish (j ) Sebastes jordani 1 1 37 SL Black rockfish (j,a) Sebastes melanops 1 1 30 SL 1 1 335 FL Bocaccio (j ) Sebastes paucispinis 20 5 21-36 SL Canary rockfish (j ) Sebastes pinniger 27 5 22-39 SL Bank rockfish (j ) Sebastes rufus 8 1 16-28 SL Stripetail rockfish (j) Sebastes saxicola 13 3 32-37 SL Hexagrammidae Lingcod (j) Ophiodon elongatus 20 9 76-81 FL Anoplopomatidae Sablefish (j ) Anoplopoma fimbria 182 14 55-136 FL 4 2 173-241 FL continued 32 Fishery Bulletin 102(1) Table 1 (continued) Class and Family Common name Scientific name June (84 stations) August 179 stations) Number F.O. Size range Number F.O. Size range Cottidae Irish lord Ij) Hemilepidotus spp. 2 1 38-40 FL Cabezon (j ) Scorpeanichthys marmoratus 12 7 33-38 SL Pacific staghorn sculpin Leptocottus armatus 1 1 180 TL Agonidae Sturgeon poacher (j) Podothecus acipenserinus 1 1 80 TL Cyclopteridae Pacific spiny lumpsucker Eumierotremus orbis 1 1 253 TL Carangidae Jack mackerel Trachurus symmetricus 111 3 364-583 FL 839 20 227-589 FL Bramidae Pacific pomfret Brama japonica 5 2 387-434 FL Anarhichadidae Wolf-eel (j) Anarrhichthys ocellatus 15 13 215-555 TL 8 7 442-582 TL Ammodytidae Pacific sandlance Ammodytes hexapterus 4 4 45-82 SL Zaprodidae Prowfish (j) Zaprora silenus 1 1 68 SL Scombridae Chub mackerel Scomber japonicus 74 6 266-421 FL Centrolophidae Medusafish Icichthys lockingtoni 3 3 37-50 SL 8 6 87-129 FL Bothidae Sanddabs (j) Citharichthys spp. 23 13 35-43 SL 3 2 269-288 TL Pacific sanddab (j ) Citharichthys sordidus 32 4 32^4 SL Speckled sanddab (j ) Citharichthys stigmaeus 60 10 30-43 SL Pleuronectidae Dover sole (j) Microstomas pacificus 2 2 40-50 SL 3 1 27-34 SL Sand sole (j) Psettichthys melanostictus 3 3 22-39 SL Slender sole (j) Eopsetta exilis 1 1 66 SL Starry flounder Platichthys stellatus 2 1 349-399 TL Curlfin sole (j) Pleuronichthys decurrens 5 3 25-31 SL English sole Parophrys vetulus 1 1 303 TL Rex sole (j ) Errex zachirus 581 12 34-79 SL 48 11 44-70 SL Molidae Ocean sunfish Mola mola 1 1 620 TL Total 5974 12,878 about one half of the nonjacks caught north of Cape Blanco in August were hatchery fish. Two factors, however, may have lead to inaccuracies in estimation of hatchery-wild ratios of coho salmon in the GLOBEC study area. First, because of low sample sizes, the data were pooled from both June and August catches for the genetic stock analysis; therefore we do not know the proportional contributions of the different release areas to the catches in either month alone. Second, all the fish released from Klamath River and Trinity River hatcheries had been clipped on the maxillary. We were unaware that the maxillary clip was being used, did not look for it, and consequently may have classified fish with this mark as unmarked. Therefore, the proportion of hatchery fish in the catch of coho salmon during GLOBEC may have been higher than is shown in Table 3. Age and growth of juvenile coho salmon Forty-three percent (24 of 56) of the juvenile coho salmon caught during the August GLOBEC cruise were preco- cious males ("jacks") according to the testes-weight to body-weight criteria of Pearcy and Fisher ( 1988). This is a much higher percentage of jacks than found among juve- nile fish caught in September 2000 in the plume study off Oregon and Washington, where only 4.5% offish (6 of 132) were precocious males or females according to the same criteria. Because the jacks were considerably larger than the nonjacks, average growth rates of the two groups were reported separately. Estimated average growth rates in FL between ocean entry and capture were higher for fish caught in the August 2000 GLOBEC cruises (1.56-2.22 mm/d) than for fish caught in any other cruises (Table 3). The fish caught in August 2000 were also larger when they entered the ocean (average 170- 178 mm FL) than fish caught in other cruises (averagel54-160 mm FL). Average growth rate of jacks from north of Cape Blanco (2.22 mm/d), was significantly higher (/-test, P<0.05) than growth rates of nonjacks (1.56-1.67 mm/d). Growth rates of nonjacks north and south of Cape Blanco were not significantly different la- test, P<0.05). The combination of large size at ocean entry Brodeur et al.: Distribution, growth, condition, origin, and associations of juvenile salmonids 33 45.0 44.5 44.0 43.5 43.0 42.5 42.0 41.5 Newport Chinook > 1 to 5 6 to 150 Coho A 1 10 5 A 6 to 150 J Oregon r California 45.0 44.5 44.0 43.5 43.0 42.5 42.0 41.5 125.5 125.0 124.5 124,0 123.5 125.5 Longitude (W) 125.0 124.5 124.0 123.5 Figure 3 Catch distribution for juvenile coho (Oncorhynchus kisutch) and chinook salmon (O. tshawytscha) for the (A) June and (B) August cruise overlaid on surface temperature contours. Plus signs are stations sampled where no salmon were caught. and favorable conditions for growth in the ocean probably contributed to the very high percentage of jack coho salmon in August 2000 in the GLOBEC study area. Estimated average growth rates between ocean entry and capture of juvenile coho salmon were higher in the GLOBEC area than in the plume study area U-tests, P<0.05). For fish caught in June, average growth rate was 1.06 mm/d and 0.63 mm/d in the GLOBEC and plume study areas, respectively. For fish caught in August or September, average growth rate was 1.57-2.22 mm/d in the GLOBEC study area and 1.17 mm/d plume in the study area (Table 3). The higher growth rates of coho salmon caught in the GLOBEC study area suggests that in 2000 conditions for growth were bet- ter there than those in the plume study area farther north off Oregon and Washington. Average instantaneous growth rates in weight were also higher (/-tests, P<0.05) for the fish caught in the June and August 2000 GLOBEC cruises (2.0 and 2.1-2.8% body wt/d, respectively) than for the fish caught in the June and September 2000 plume study cruises (1.2 and 1.7 % body wt/d, respectively; Table 4A). In addition, the average condition index (CI) of juve- nile coho salmon in June was significantly higher (/-test, P=0.03) in the GLOBEC study area (1.12, n=32, SD=0.087) than in the plume study area (1.07, n=245, SD=0.117). Similarly, the average CI of nonjack juvenile coho salmon was higher (/-test, P=0.002) in August in the GLOBEC study area (1.24, n=32, SD=0.096) than in September in the plume study area (1.18, n=132, SD=0.100). Both the high instantaneous growth rates in weight and the high CI of juvenile coho salmon caught in the GLOBEC study area suggest that conditions for growth of coho salmon in this area were very good in 2000. Growth rates estimated from the few CWT fish caught during these cruises (Table 4B) were similar to, and help validate, the growth rates estimated from scales (Table 4A). Average weights at time of ocean entry back-calculated from scales for coho salmon caught in June in the GLOBEC area and in all months in the plume study area (Table 4A) were slightly higher than the average weights of hatchery coho salmon at time of release (Appendix Table 1). For ex- ample, in the plume study area, average back calculated weights at ocean entry ranged from 37.5 g to 42.4 g (Table 4A) — slightly higher than the expected average weights at release of about 32-33 g based on the stock composi- 34 Fishery Bulletin 102(1) Table 2 Summary of mean, standard deviation, and range of FL measured in the field, weight measured in the laboratory, and condition index (CI) of subyearling (age 0.0) and yearling (age 1.0) chinook salmon and yearling (age 1.0) coho salmon caught during the June and August cruises north (N) and south (S) of Cape Blanco (latitude 42.837°). Precocious coho salmon are indicated with a "J". Field FL (mm) Laboratory weight (g) C.I. (wtx 10 5 / FL 3 ) n Mean SD Range Mean SD Range Mean SD Chinook (age 0.0) June (N) 1 121 — — 18 — — 1.04 — August (N) 1 172 — — 70 — — 1.37 — August (S) 125 134 12 109-175 28 9 12-70 1.10 0.08 Chinook (age 1.0) June (N) 27 229 42 144-280 178 91 33-306 1.32 0.10 June(S) 1 174 — — 67 — — 1.28 — August (N) 54 229 26 187-318 164 72 80-468 1.32 0.09 August (S) 35 231 35 190-349 176 94 80-535 1.32 0.07 Coho (age 1.0) June (N) 30 161 33 122-276 56 51 19-292 1.13 0.08 June (S) 2 172 172-172 49 1 48-49 0.95 0.01 August (N-J) 24 365 31 310-415 690 209 375-1198 1.38 0.12 August (N) 24 285 51 210-385 326 188 97-766 1.26 0.10 August (S) 8 293 33 239-334 308 103 157-433 1.19 0.05 Table 3 Catch, percentage of the catch that was marked, estimated percentage of hatchery origin, size of scale sample, FL at ocean entry (OE) back calculated from scales, FL at capture, and estimated growth rate in FL while in the ocean for juvenile coho salmon caught during the 2000 GLOBEC and Columbia River plume studies. All length data are from the scale sample only. An ocean entry date of 15 May was used when calculating growth rate in FL. Cruise Catch (n) Marked Estimated % Scale sample hatchery origin (n) Back- calculated FL at OE (mm) mean (SD) FL at capture (mm) Growth rate (mm/d) mean (SD) mean(SD) GLOBEC June 2000 32 32% 33% 11 155 (29.0) 177(42.3) 1.06(1.01) Aug 2000 North of C.Blanco Jacks 24 71% 74% 19 170(22.8) 370(28.1) 2.22 (0.35) Nonjacks 24 46% 48% 9 178(21.6) 309(46.1) 1.67 (0.51) South of C. Blanco Nonjacks 8 38% 39% 6 178(13.0) 303 (29.3) 1.56 (0.22) Plume study May 2000 165 68% 76-80% ; 79 157(16.5) 166(17.7) 0.97(1.15) Jun 2000 245 76% 90% 97 160(14.5) 185(23.4) 0.63 (0.53) Sep 2000 132 65% 75% 76 154(19.0) 305 (24.9) 1.17(0.23) ' No genetic stock analysis was available. The higher estimate assumes the same stock composition as in June, hatchery fish were from the Columbia River. the lower estimate assumes that all Brodeur et al.: Distribution, growth, condition, origin, and associations of |uvenile salmonids 35 A 0.004 0002 -0.002 -0.004 □ Wet Wt (Somatic Growth) to as -0.006 "D Dry Wl (Energetic Growth) CO 1) 0.02 -, o B 0.01 - — L — H^H -0.01 - -0.02 - -0.03 - -0.04 - -0.05 - l -0.06 - -0.07 - Cape Blanco Cape Blanco North South Figure 4 Wet and dry weight residuals ( + 1 standard error) for (A) yearling chinook (On- corhynchus tshawytscha) and (B) juvenile coho salmon (O. kisutch) collected North and South of Cape Blanco. Weight residuals are derived from the linear relationship between fork length and wet or dry weight (log-transformed data) of juvenile salmon captured in June and August. tion of these catches (Teel et al., 2003) and the release weights (Appendix Table 1). Similarly, the back-calculated weight at ocean entry in June in the GLOBEC area (45.5 g) was slightly higher than the expected average weight at hatchery release (about 41 gl based on the stock compo- sition (Table 5) and the average release weights. These fairly small differences between back-calculated size at ocean entry and average size at release could be due to growth during downstream migration, selectively higher mortality of small smolts, or a bias in the back-calculation procedure. However, the average back-calculated weights at time of ocean entry offish caught in August in the GLOBEC study area (60-69 g) were over two standard deviations above the average weights of hatchery fish released from the Oregon coast or northern California — the main contributors to this catch (Appendix Table 1). These were obviously atypical coho salmon, and the very high proportion of jacks (preco- 36 Fishery Bulletin 102(1) Table 4 (A) Weights at ocean entry I OE ) back-calculated from scales, weights at capture and estimated instantaneous rates of growth while in the ocean iGl for juvenile coho salmon caught during the 2000 GLOBEC and Col umbia River plume studies. An ocean entry date of 15 May was used when calculating growth rate. (B) Similar data for CWT fish. Growth rates of the CWT coho salmon were estimated for the periods between hatchery release and capture in the ocean. A Cruise ;; Back-calc. Wt. at OE (g) Weight at capture (g) G mean (SD) mean(SD) mean (SD) GLOBEC June 2000 11 45.5 (26.8) 78.0(76.4) 0.020(0.015) Aug 2000 North of C. Blanco Jacks 19 68.9(27.2) 719.7(200.0) 0.028 (0.005) Nonjacks 9 59.5 (26.3) 419.2(177.2) 0.023 (0.006) South of C. Blanco Nonjacks 6 60.3(12.8) 336.2 (96.2) 0.021 (0.002) Plume study May 2000 79 39.4 (10.8) 47.9(14.6) 0.020(0.024) Jun 2000 97 42.4(12.5) 71.9(33.3) 0.012(0.009) Sep 2000 75 37.5(13.7) 347.2(158.3) 0.017(0.003) B Cruise n Wt. at release (g) Wt. at capture (g) G mean (SD) mean (SD) mean (SD) GLOBEC Jun 2000 4 44.4(1.3) 86.6 (30.9) 0.018(0.005) Aug 2000 3 35.6 (9.8) 395.7(215.0) 0.024(0.003) Plume study Jun 2000 11 28.3(4.5) 66.1(32.3) 0.012(0.005) Sep 2000 10 33.4(10.91 392.4(283.3) 0.018(0.002) cious, sexually developed males) among the fish was prob- ably a consequence of their very large size at ocean entry and their high rates of growth in the ocean. Freshwater origins of juvenile salmonids Allozyme data were collected from samples of 247 chinook salmon, 88 coho salmon, and 58 steelhead. Genetic mixed stock analyses indicated that chinook salmon in June were predominately (54%, SD=0.18) from rivers and hatcheries along the mid Oregon coast, an area immediately north of Cape Blanco (Table 5, Fig. 5). In August, chinook salmon were largely from rivers that enter the sea south of Cape Blanco. Fish from the Sacramento and San Joaquin rivers in northern California were estimated to comprise 90% (SD=0.07) of the chinook salmon sampled in August north of Cape Blanco. The largest concentration of chinook salmon we sampled was south of Cape Blanco in August, and these fish were mostly from rivers in southern Oregon (539(, SD=0.10) and the Sacramento and San Joaquin rivers (20%, SD=0.05). Chinook salmon from the Colum- bia River Basin were also present, but were estimated to comprise only 18% (SD=0.15) of the June sample and 8% (SD=0.05) of the August sample north of Cape Blanco. Recoveries of hatchery chinook salmon bearing coded-wire tags (CWT) provided direct evidence of stock origins for ten fish, all taken in trawls north of Cape Blanco (Table 5). These data reveal that hatchery fish released from the Umpqua River on the central Oregon coast (;;=6), Columbia River Basin («=3) and Sacramento River (« = 1) contributed to our sample of chinook salmon. The propor- tion of CWT fish from the Umpqua River in our August catch north of Cape Blanco (8%) indicated that the con- tribution of mid Oregon coastal fish was underestimated in the genetic analysis likely because of the small size of the mixture sample. Genetic estimates of coho salmon indicated that most fish originated from coastal Oregon rivers north of Cape Blanco (479S , SD=0.10) and from the Columbia River (13%, SD=0.08 ) (Table 5 ). However, a substantial proportion (40 r /i , SD=0.09) of coho salmon were from coastal rivers south of Cape Blanco, a region that includes spawning populations in the Rogue and Klamath rivers. Eight coho salmon in our sample contained CWTs and showed that fish from Brodeur et al.: Distribution, growth, condition, origin, and associations of juvenile salmonids 37 Table 5 Estimated percentage stock compositions, samples sizes, and recoveries of coded wire tags (CWTs) for chinook and coho salmon and steelhead sampled in trawl surveys along the Oregon and California coasts in 2000 Some of tht major baseline stocks are given for coastal stock groups. None of the steelhead sampled contained coded wire tags. June (rc=35) August (?!=157) August (n=55) Entire South of North of Study Area Cape Blanco Cape Blanco Chinook salmon stock group Est. SD CWT Est. SD CWT Est. SD CWT Columbia and Snake Rivers 18 0.15 2 3 0.03 8 0.05 1 North Oregon coast (Nehalem, Trask, Alsea, and Siuslaw Rivers) 0.00 0.00 0.00 Mid Oregon coast (Umpqua, Coquille, Sixes, and Elk Rivers) 54 0.18 3 3 0.03 1 0.02 3 South Oregon coast (Rogue. Chetco, and Winchuck Rivers) 26 0.16 53 0.10 0.00 Klamath and Trinity Rivers 0.00 14 0.07 0.00 North California Coast (Mad, Eel, and Mattole Rivers) 2 0.05 7 0.07 1 0.04 Sacramento and San Joaquin Rivers 0.00 20 0.05 90 0.07 1 June and August (rc=88) Coho salmon stock group Entire study area Est. SD CWT Columbia River 13 0.08 2 North and Mid Oregon coast (Nehalem, Siletz, Alsea, Umpqua, and Coos Rivers) 47 0.10 5 Rogue and Klamath Rivers 40 0.09 1 North California Coast (Mad, Russian, Little, and Scott Rivers) 0.00 June and August (n=58) Steelhead trout stock group Entire study area Est. SD Columbia and Snake Rivers 0.00 North and Mid Oregon coast (Nehalem, Siletz, Alsea, Umpqua, Coos, and Coquille Rivers) 1 0.03 South Oregon coast (Elk, Rogue, Chetco, and Winchuck Rivers) 53 0.08 Smith, Klamath, and Trinity Rivers 0.00 North California Coast (Mad, Eel, and Ten Mile Rivers) 10 0.05 Sacramento and San Joaquin Rivers 14 0.05 Central and South California Coast (San Lorenzo River and Scott, Pauma, and Gaviota Creeks 3 0.02 Unknown 19 — hatcheries in the Umpqua River (n=5), Rogue River (n=l), and Columbia River (n=2) were in our study area. Genetic analysis of steelhead samples showed that a large proportion were from the Rogue River and nearby coastal streams (53%, SD=0.08). Steelhead from the Sacra- mento and San Joaquin rivers (14%, SD=0.05) and north- ern California coastal rivers (10%, SD=0.05) were also present. Estimates for steelhead originating from rivers north of Cape Blanco and from south of the San Francisco Bay were near zero. Approximately 19% of the steelhead mixture was not allocated to any source population, sug- gesting that our baseline data for the species is incomplete. No steelhead in our collections contained CWTs. Species associations of juvenile salmonids and other species From cluster analysis of species based on station assem- blages (Fig. 6), MRPP of both sample periods showed strong within-group agreement (P<0.0001) at the first level (two groups); all subsequent groups had sequentially higher levels of within-group agreement. As a result, the cutoff level was determined by balancing a lower percent infor- mation remaining (<30%) in the model while retaining bio- logically meaningful groups. For June this cutoff resided at the second level (three groups) and for August, at the third level (four groups ). For the June cruise, all salmonids includ- 38 Fishery Bulletin 102(1) 1 A I 127° i 122" 1 117"W — 50°N Vancouver "~-~~ Island ^fc--- B.C. - Pacific Ocean Olympic Peninsula Puge! ^ Sound , r" r£*\ •/ - 46" Columbia R -5sk^ "X Columbia R L wA J Snake R _ — 42" N -38" ^ Newport Cape Blanco / ,-.-, / • Yj Crescent City vC7 Eel R. /\\ I s" Umpqua Rogue R CU / o> r /, 3 ) A 3 ( L- o \ ;o V i R. 1 OR V ID CA 3 arvJ u 1 1 1 1 00 200 km I B 1 1 127- 122- 1 117"W June -so-N entire study area .^^ - © f|°oo ~ 46 ' August north of Cape Blanco ° o o 7\ - 4 -' W i August south of #•'-. i Cape Blanco i # • 100 200 km 1 1 •• 1 1 — 50* N c — r 132" 127° 122° ' 46 'June and August entire study area — 42" O • N J_ _L J. Figure 5 (A) Map of study area and location of GLOBEC sampling (hatching). (B) Stock compositions of chinook salmon (Oncorhynchus tshawytscha). Stock groups are North of Columbia River (grey), Columbia River Basin (green), north Oregon coast (pink), mid Oregon coast (yellow), south Oregon coast (dark blue), Klamath River Basin (black), north California coast (light blue), and Central Valley (red). (C) Stock compositions of coho salmon (O. kisutch). Stock groups are Columbia River (green), mid and north Oregon coast (dark pink), Rogue and Klamath rivers (blue), and north California coast (orange). Brodeur et al.: Distribution, growth, condition, origin, and associations of juvenile salmonids 39 June 100 Information remaining (%) 75 50 25 H 1 1 1 H Coho Chinook (a) Wolfeel Chinook (j) Lmgcod Steelhead Sablefish Market squid Whitebait smelt Pacific herring Surf smelt Darkblotched rockfish — , Yellowtail rockfish — T~ Rex sole — Speckled sanddab — i August 100 r- Information remaining (%) 75 50 25 H 1 1 1 h i Coho (a) Coho (j) Chinook (a) Chinook (j) Surf smelt Steelhead Medusafish Pacific saury Wolfeel Osmeriid (j) Blue shark Northern anchovy Rex sole Chub mackerel Pacific sardine Jack mackerel Figure 6 Cluster species groupings by cruise. The dashed lines indicate the cutoff levels for each cluster group. See Table 1 for scientific names. i> ing steelhead were classified within the same grouping that included several pelagic juvenile taxa, including wolf-eel, lingcod, and sablefish (Fig. 61. Two other clusters that were not associated with juvenile salmon included a southern inshore group consisting of market squid. Pacific herring, and two species of smelt and an offshore northern group consisting primarily of juvenile rockfish and rex sole. For the August cruise, all salmonid juveniles and adults again clustered together in one large group with surf smelt and medusafish ( Fig. 6 ). The remaining three groups were much smaller and consisted primarily of offshore pelagic species. Cluster analysis of stations based on species assem- blages, and subsequent examination of the cutoff level us- ing MRPP, resulted in three groupings from both sample periods (Fig. 7). MRPP revealed strong within-group agreement for all levels (P<0.0001); however, delineation at three groups was based on maintaining lower percent in- formation remaining (<30%) and still having a meaningful level of resolution. There was some measure of geographic separation among the three groups (Fig. 7). In June, group A was predominantly inshore and mostly in the southern half of the sampling area, group B was found mainly in the middle shelf region and was more northern, and group C was found predominantly offshore. In August, group A consisted of only three stations, all south of Cape Blanco, whereas groups B and C both spanned the entire shelf and offshore region and had no particular north-south affin- ity (Fig. 7). ISA of the groups from both sampling periods showed that only groups A and C had indicator species (Tables 6 and 7), whereas the intermediate groups had none. Ordination analyses and environmental correlates NMS ordination of the June sampling period (Fig. 8A) revealed most of the variance in the data: axes 1 and 40 Fishery Bulletin 102(1) June 2000 AAAAAn] a 44.5- □ A * rw 44.0- cPa A a n nrriAAA d 43.5- n nriAA'fY/ • 43.0- A \ D AA \ ( □□ ao/ duster Groupings 42.5- ' Group A O , Group B A aA Group C [ 42.0- OnA , , , i i , , , , 1 42.0- August 2000 rr D| &A A 125.5 125.0 124.5 124.0 1235 125.5 125.0 124.5 124.0 123.5 Longitude (W) Figure 7 Map showing locations of cluster station groupings by cruise. Table 6 Indicator species analysis showing indicator values for dominant pelagic nekton captu mean, standard deviations (SD), and P- values for each cluster grouping. Cluster Group mined to be indicators of that group. •ed in pelagic trawls during June 2000 and B did not have any species that were deter- Group Species Observed indicator value (IV) Indicator value IV from randomized groups P-value Mean SD A chinook (age 0.0 1 61.0 15.7 6.54 <0.001 A lingcod 26.1 12.6 5.67 0.024 A Pacific herring 71.7 12.8 5.88 <0.001 A surf smelt 86.5 11.8 5.59 <0.001 A whitebait smelt 31.5 10.4 5.55 0.007 A market squid 50.8 15.0 6.20 <0.001 C darkblotched rockfish 66.8 1 5 8 6.31 <0.001 C rex sole 46.0 15.0 6.24 0.002 C sablefish 31.1 16.2 6.32 0.035 C speckled sanddab 52.5 13.4 5.94 0.001 C yellowtail rockfish 98.8 19.0 6.30 <0.001 3 represented 31% and 237f, respectively (stress=16.3). Temperature, depth and salinity best explained the ordi- n;it ion of stations, representing a cross shelf gradient from nearshore high levels of salinity to increasing temperature and depth offshore. Ordination of August stations (Fig. 8B) represented 42' i of the variance in the data, and 23% of the variance was loaded on axis 2 and 19% on axis 3 (stress=19.4). As with June, salinity increased toward the coast and temperature and depth increased off the shelf. The groups derived from the cluster analysis tended to group together in multivariate space, with the exception of group B in the June cruise (triangles in Fig. 8A). Brodeur et al.: Distribution, growth, condition, origin, and associations of |uvenile salmonids 41 Table 7 Indicator Species Analysis showing indicator values for dominant pelagic nekton captured in pelagic trawls during August 2000 and mean, standard deviations (SD), and P-values for each cluster grouping. Cluster Group B did not have any species that were determined to be indicators of that group. Group Species Observed indicator value (IV) Indicator value IV from randomized groups P-value Mean SD A chinook (age 1.0) 76.5 21.3 11.18 0.004 A A chinook (age 0.0) surf smelt 80.4 97.9 22.1 12.4 11.62 8.21 0.003 <0.001 C chub mackerel 33.3 12.8 8.88 0.021 C jack mackerel 73.7 23.0 11.86 0.006 Table 8 Results of statistical tests for habitat associations between juvenile salmon and environmental or station variables from each cruise in 2000. Fish marked by zeros indicate subyearlings and those marked with one indicate yearlings. Shown are the P-levels for 5000 randomizations of the cumulative frequency of the habitat variable and the proportion of the standardized salmon catch associated with each habitat observation. Results are based on the Cramer von-Mises test statistic determined from binned data for depth and neuston biomass. Significance values <0.05 are shown in boldface. Cruise Jun Aug Taxon and age Surface temp. Surface salinity 1-m chlorophyll Bottom depth Neuston biomass chinook (age 1.0) 0.30 0.60 0.13 0.18 0.13 coho (age 1.0) 0.33 0.48 0.21 0.17 0.31 chinook (age 0.0) 0.36 0.25 0.13 0.35 0.42 chinook (age 1.0) 0.04 <0.01 <0.01 0.02 0.29 coho (age 1.0) 0.68 0.04 0.07 0.02 0.45 There were few instances where the habitat associations of juvenile salmon were significantly different from the distribution of environmental variables sampled (Table 8). None of the variables were significant for yearling chinook and coho salmon in the June sampling (no subyearling salmon were caught during that cruise). In August, all the variables except neuston biomass were significant for yearling chinook salmon. These fish were collected at cooler temperatures, higher salinities, higher chlorophyll-o con- centrations, and at shallower depths than have been typi- cally recorded (Fig. 9). Coho salmonjuveniles were found in higher salinities and shallower depths than at the sampled habitat (Fig. 9). These results correlated with the capture of juvenile chinook salmon and to a lesser with extent coho salmon at nearshore stations in the upwelling zone. Discussion Understanding the mechanisms underlying the dynamics of multispecies communities is one of the biggest challenges in ecology. Most communities contain many interacting spe- cies, each of which is likely to be affected by multiple biotic and abiotic factors. In order to effectively characterize a system, we need to differentiate variability resulting from both temporal and spatial factors. Our observations took place during two time periods of about 20 days each and thus were not synoptic "snapshots" of the system. Indeed, during our June sampling, conditions changed markedly from the beginning to the end of the cruise because of the arrival of an anomalous major southwest storm ( Batch- elder et al., 2002), which likely completely altered the hydrography and biology of the system. Thus, short-term temporal variability may obscure patterns observed over the spatial scale of our sampling. The pelagic nekton community sampled during these cruises was not that different from what had previously been shown for purse seine and trawling collections off the coast of Oregon and Washington ( Brodeur and Pearcy, 1986; Emmett and Brodeur, 2000; Brodeur et al., 2003). The early summer nekton community was dominated by coastal forage fishes such as smelts and Pacific herring, but also comprised juveniles of many rockfish, sculpin, and flatfish species. These winter-spring spawning species eventually settle out to demersal habitats sometime in summer (Shenker, 1988; Doyle, 1992), which may in part explain the paucity of these taxa in the August cruise. In contrast, the August nekton community consisted of large, 42 Fishery Bulletin 102(1) highly migratory species such as Pacific sardines, jack mackerel, and chub mackerel. Pacific sardine, which was almost completely absent from the system in the 1980s, has undergone a substantial resurgence and is now one of the most abundant species off the coast in summer (Brodeur et al., 2000; Emmett and Brodeur, 2000; McFarlane and Beamish, 2001). It should be noted, however, that some of the differences between cruises could be accounted for by the inclusion of substantially more offshore stations during A; a a REXS SPSD cr A A A 3 A U STHD t A MASO o WBSM DBRF °-,YTRF D cP SABF A Temperature Depth , Salinity n n LGCD - <% PHER »* D d 1 CHIN1 COHO A A * 1 Axisl (r 2 =0.31) CO CO X < B A A * A * REXS A D STHD A COHOA A D a coftoj D A a D -Depth a Salinity D Temperature CHINO BLSH OSMJ ° CO CD PSAR CHIN1 O o a o ^OEL a °0 CO o Axis 2 (r 2 =0.23) Figure 8 Nonmetric multidimensional scaling (NMS) ordination plot of stations and nekton species with environmental parameters from June (A) and August (B) 2000 GLOBEC cruises. Station symbols denote: onshore tO>. mid-shelf !▲). and slope (D) groupings; Species abbreviations denote the following taxa: CHIN (chinook, age 0), CHIN 1 (chinook, age al.ll, STHD (steelhead trout). SUSM (surf smelt), PSAU (Pacific saury), WOEL (wolf-eel juvenile), OSM J (osmerid juvenile), REXS (rex sole, larval i, MEDF (medusafish ), PSAR (Pacific sardine), .JAMA (jack mack- erel), CHMA (chub mackerel), NANC (northern anchovy). BLSH (blue shark). the second cruise. Our results from the community analy- ses suggest that juvenile salmon tend to co-occur with each other and with a variety of other pelagic nekton, including adult salmon, and that this spatial overlap varies tempo- rally. Brodeur et al. (2003), in analyzing community struc- ture based on previous pelagic sampling data off Oregon and Washington, arrived at similar results. In both studies, the geographic boundaries of the pelagic assemblages often overlap and are not as distinct as demersal assemblages. However, the pelagic environment is much more spatially and temporally heterogeneous than the demersal environ- ment, and many of the species examined in our study are highly mobile and are likely to respond quickly to changing conditions. Research is presently underway to examine the trophic interactions among salmonids and with other sym- patric nekton species in order to determine what ecological relationships (e.g. predation, competition), if any, are occur- ring in this system. From the results of our sampling, we concluded that ju- venile salmonids, with the possible exception of steelhead, occupy the cool, high salinity, inshore upwelling regions off the southern Oregon coast. However, particularly for the June cruise, many of the coho and chinook salmon juveniles collected may have recently entered the ocean with little time to disperse offshore, so that the capture location may not reflect true habitat preferences. Moreover, the vertical dimensions of our trawl also precluded us from sampling the nearshore, subtidal regions where some subyearling chinook may reside shortly after entering the ocean. Salmon and steelhead differed considerably in stock com- position. The pattern for coho salmon was similar to that of chinook salmon in that fish from sources both north and south of Cape Blanco contributed to our catches. However, steelhead from rivers north of Cape Blanco were absent, presumably having migrated offshore and north shortly after entering the sea, as shown by Pearcy et al. (1990). Although our stock composition estimates for steelhead should be viewed with caution because of an incomplete ge- netic baseline and a relatively small number of samples, our findings support Pearcy et al.'s suggestion that steelhead from rivers south of Cape Blanco have a unique marine distribution and reside throughout the summer in the up- welling zone off northern California and southern Oregon. Our study revealed seasonal shifts in the abundance and stock composition of juvenile salmonids. Although salmo- nids comprised small portions of the vertebrate catches of both the June and August cruises, juvenile chinook salmon, coho salmon, and steelhead were much more abundant later in the summer, likely indicating that fish moving into our study area are from shoreline or riverine habitats. The greater abundance of chinook salmon in late summer can be explained in part by the northern migration offish that originated in rivers south of our study area. Chinook salmon from the Sacramento and San Joaquin rivers in California's Central Valley comprised substantial propor- tions in the August catches both south (20%) and in nth i 90' i ) of Cape Blanco. In contrast, the few chinook salmon caught in June were mostly (549r ) from streams that en- ter the sea immediately north of Cape Blanco such as the Umpqua, Coquille, Sixes, and Elk rivers. Chinook salmon Brodeur et al.: Distribution, growth, condition, origin, and associations of juvenile salmonids 43 E o •Chinook 1 Coho 1 .0 -Habitat 12 14 Water temperature (C) j ■*" 09 - ~~, r. ..---' OR - , * 07 - / ■" . .. 06 - r- ' 05 - r' 04 - 03 - - -Chinook 1 02 - f J " Coho 1 n 1 - 1 V Habitat n - 10 15 Chla concentration 1 09 08 07 06 0.5 04 03 0.2 1 31.50 J i - - -Chinook 1 X ' Coho 1.0 Habitat Y 1 ^ } # > ,' 4 ja _ _ _ J 3250 3300 Salinity (PSU) •Chinook 1 -Coho 1.0 -Habitat 100 150 200 Water depth (m) Figure 9 Cumulative distribution curves for salmon and environmental or station variables. Only the August variables that showed at least one significant difference are included. See Table 8 for results of the statistical tests. from these rivers are known to primarily migrate north of our study area along the coast (Nicholas and Hankin, 1988). By August, fish from these stocks were nearly absent from our samples. Oregon rivers south of Cape Blanco, an area that includes the Rogue, Chetco, and Winchuck riv- ers, produce chinook salmon with a more southerly pattern of ocean migration (Nicholas and Hankin, 1988; Myers et al., 1998). Chinook salmon from these rivers were found throughout the summer and contributed 53% to our largest catches of chinook salmon along transects south of Cape Blanco in August. Results from our 2000 GLOBEC cruises identified Cape Blanco as an important breakpoint in salmonid life-his- tory variation. Stock distributions of both juvenile salmon and steelhead indicated that different migration patterns of fish originating from southern and northern rivers are evident during their early marine phase. Our August sur- vey also revealed sharp contrasts in life-history type and freshwater origin between the juvenile chinook salmon population in the marine area north of Cape Blanco and that to the south. Chinook salmon captured north of Cape Blanco were nearly all yearlings and largely from the Sac- ramento River drainage. Subyearlings predominated in our catches south of Cape Blanco and included a much larger proportion offish from coastal streams in southern Oregon and northern California. Comparisons of our results with similar studies conduct- ed further north show differences in salmonid migrations on a somewhat broader geographic scale. In several years of sampling during the summers of 1981 through 1985 off the central Oregon to northern Washington coast, most juvenile chinook salmon bearing CWTs were from Columbia River hatcheries (Pearcy and Fisher, 1990; Fisher and Pearcy, 1995). Only one tagged chinook salmon from a river south of Cape Blanco (Klamath River) was captured. Pearcy and Fisher also found that juvenile coho salmon were largely from the Columbia River and that smaller contributions were from coastal rivers north of Cape Blanco. Their find- ings have been corroborated by more recent surveys in the same region (Emmett and Brodeur, 2000) using genetic 44 Fishery Bulletin 102(1) data (Teel et al., 2003). Samples from subsequent cruises will be used to examine the persistence of such fine- and broad-scale geographic structure in the juvenile migrations of salmonid stocks. A major source of error in our estimates of growth rates of juvenile coho salmon back-calculated from scales was uncertainty of when individual fish entered the ocean. We used a single date of ocean entry for all fish (15 May), but individual fish, of course, entered the ocean at different times over the course of a month or more. Consequently, coefficients of variation were relatively large (84—119% and 75-120% of mean growth rate in FL and weight, respec- tively) for fish caught in May and June, when errors in es- timated growth periods likely were large in relation to the actual growth periods. Conversely, coefficients of variation were relatively small ( 14-30% and 10-26% of growth rate in FL and weight, respectively) for fish caught in August or September, when errors in estimated growth periods likely were small in relation to the actual growth periods. (Note the decrease in standard deviation of mean growth rates with month of capture in Tables 3 and 4A). Growth rates of CWT coho salmon between hatchery release and capture in the ocean (Table 4B) were very similar to the growth rates of unmarked salmon estimated from scales for the same months and areas. In addition, the growth rates of the former group ( CWT coho salmon ) helped to validate the growth rates of the latter group (Table 4A). Significant differences in growth and condition of ju- venile coho salmon indicate that different oceanographic environments exist north and south of Cape Blanco. The length of the fish indicated that substantial growth oc- curred in juvenile coho salmon during the study period. As- sessment of other growth features (condition) revealed that juvenile coho salmon grew better north of Cape Blanco. Because we included measurement of condition in both the June and August period in the evaluation, changes in stock composition, described earlier, may be partly responsible for this observation. Although genetic stock composition was different between months, month of sampling was not a significant factor, suggesting that stock composition is not likely a significant factor affecting the difference in condition (a performance metric) of juvenile salmon north and south of Cape Blanco. Several lines of evidence further support the hypothesis that areas north of Cape Blanco benefit juvenile yearling chinook and coho salmon. There were greater numbers of juvenile yearling chinook and coho salmon to the north of Cape Blanco. Although our overall sampling effort was greater north of Cape Blanco, in the mesoscale portion of our survey designed to assess general distribution patterns, more yearling chinook and coho salmon were captured north of Cape Blanco. Secondly, when we evaluated the growth rate of juvenile coho salmon in the GLOBEC region compared to juveniles captured off northern Oregon and Washington, juveniles from the GLOBEC region grew much better. The similar tracking of resource (distribution and abundance) and performance (measured in terms of either somatic and energetic growth or growth rate) metrics for juvenile yearling chinook salmon and coho salmon ninth of Cape Blanco suggests that habitat quality in this region was better. The results of this study help define the biogeo- graphical zones for salmon growth and establish regional- based management strategies for depleted salmon stocks. Acknowledgments We thank the captain and crew of the FV Sea Eagle for their expert help in conducting the trawling operations under sometimes adverse weather conditions. We are grateful to Jackie Popp-Noskov, Paul Bentley, Marcia House, and Becky Baldwin for assistance in field sampling. Donald Van Doornik and David Kuligowski collected the genetic data. We thank Anne Marshall for the use of unpublished chinook salmon allele frequency data. Stephen Smith and Alex De Robertis helped with the statistical analy- sis. Earlier versions of this manuscript were improved by the helpful comments of two anonymous journal reviewers. Research was conducted as part of the U.S. GLOBEC program and was jointly funded by the National Science Foundation (Grant no. OCE-0002855) and the National Oceanic and Atmospheric Administra- tion (NOAA). We also acknowledge the Bonneville Power Administration for funding the plume study. Literature cited Aebersold, P. B., G. A. Winans, D. J. Teel. G. B. Milner. and F. M. Utter. 1987. Manual for starch gel electrophoresis: a method for the detection of genetic variation. NOAA Tech. Rep. NMFS 61, 19 p. Batchelder, H. P., J. A.Barth, M. P. Kosro, P. T. Strub, R. D. Brodeur, W. T. Peterson, C. T. Tynan. M. D. Ohman, L. W. Botsford, T. M. Powell, F. B. Schwing, D. G. Ainley. D. L. Mackas, B. M. Hickey, and S. R. Ramp. 2002. The GLOBEC Northeast Pacific California Current System program. Oceanogr. 15:36-47. Barth, J. A., S. D. Pierce, and R. L. Smith. 2000. A separating coastal upwelling jet at Cape Blanco. Oregon and its connection to the California Current System. Deep-Sea Res. 47(part II):783-810. Bradford, M. J. 1995. Comparative review of Pacific salmon survival rates. Can. J. Fish. Aquat. Sci. 52:1327-1338. Brodeur, R. D.. P. J. Bentley, R. L. Emmett. T. Miller, and W. T Peterson. 2000. Sardines in the ecosystem of the Pacific Northwest. Proc. sardine 2000 symposium. Pacific States Marine Fish. Comm., p. 90-102. [Available from Pacific States Marine Fisheries Commission, Portland, OR. I Brodeur, R. D., J .P. Fisher. Y. Ueno, K. Nagasawa, and W. G. Pearcy. 2003. An east-west comparison of the Transition Zone coastal pelagic nekton of the North Pacific Ocean. J. Oceanog. 59(41:415-434. Brodeur, R. D., H. V. Lorz, and W. G. Pearcy. 1987. Food habits and dietary variability of pelagic nekton off Oregon and Washington. 1979-1984. NOAA Tech. Rep. NMFS 57, 32 p. Brodeur. R. D., and W. G. Pearcy. 1986. Distribution and relative abundance of pelagic non- Brodeur et al.: Distribution, growth, condition, origin, and associations of juvenile salmonids 45 salmonid nekton off Oregon and Washington, 1979-1984. NOAA Tech. Rep. NMFS 46, 85 p. 1992. Effects of environmental variability on trophic inter- actions and food web structure in a pelagic upwelling ecosystem. Mar. Ecol. Prog. Sen 84:101-119. Busby, P. J., T. C. Wainwright, G. J. Bryant, L. Lierheimer, R. S. Waples, F. W. Waknitz, and I. V. Lagomarsino. 1996. Status review of west coast steelhead from Washing- ton, Idaho, Oregon, and California. NOAA Tech. Memo. NMFS-NWFSC-27, 261 p. Dawley, E. M., R. D. Ledgerwood, and A. Jensen. 1985. Beach and purse seine sampling of juvenile salmonids in the Columbia River estuary and ocean plume, 1977-1983. Vol. 1, Procedures, sampling effort, and catch data. NOAA Tech. Memo. NMFS F/NWC-74, 260 p. Debevec, E. M., R. B. Gates, M. Masuda. J. Pella, J. Reynolds, and L. W. Seeb. 2000. SPAM (Version 3.2): Statistics program for analyzing mixtures. J. Hered. 91:509-511. Doyle, M. J. 1992. Neustonic ichthyoplankton in the northern region of the California Current ecosystem. Calif. Coop. Oceanic Fish. Invest. Rep. 33:141-161. Emmett, R. L., and R. D. Brodeur. 2000. The relationship between recent changes in the pelagic nekton community off Oregon and Washington and physical oceanographic conditions. Bull. North Pac. Anadr. Fish. Comm. 2:11-20. Fisher, J. P., and W. G. Pearcy. 1988. Growth of juvenile coho salmon (Oncorhynchus kisutch) off Oregon and Washington, USA in years of dif- fering coastal upwelling. Can. J. Fish. Aquat. Sci. 45: 1036-1044. 1995. Distribution, migration, and growth of juvenile Chi- nook salmon, Oncohychus tshawytscha, off Oregon and Washington. Fish. Bull. 93:274-289. Francis, R. C, and S. R. Hare. 1994. Decadal-scale regime shifts in large marine ecosys- tems of the Northeast Pacific: a case for historical science. Fish. Oceanogr. 3:279-291. Friedland, K. D., and R. E. Haas. 1996. Marine post-smolt growth and age at maturity of Atlantic salmon. J. Fish Biol. 48:1-15. Gauch, H. G.Jr. 1982. Multivariate analysis in community ecology. Cam- bridge Univ. Press, N.Y. 298 p. Hare, S. R., N. J. Mantua, and R. C. Francis. 1999. Inverse production regimes: Alaska and West Coast salmon. Fisheries 24:6-14. Hinch, S. G, M. C. Healey, R. E. Diewert, K. A.Thomson, R. Hourston, M. A. Henderson, and F. Juanes. 1995. Potential effects of climate change on marine growth and survival of Fraser River sockeye salmon. Can. J. Fish. Aquat. Sci. 52:2651-2659. Jakob, E. M., S. D. Marshall, and G. W. Uetz. 1996. Estimating fitness: a comparison of body condition indices. Oikos 77: 61-67. Kope, R.. and T. Wainwright. 1998. Trends in the status of Pacific salmon populations in Washington, Oregon, California, and Idaho. North Pac. Anadr. Fish. Comm. Bull. 1:1-12. Kruskal, J. B. 1964. Nonmetric multidimensional scaling: a numerical method. Psychometricka 29:115-129. Lawson, P. W. 1993. Cycles in ocean productivity, trends in habitat quality, and the restoration of salmon runs in Oregon. Fisheries 18:6-10. Lorenzen, K. 1996. The relationship between body weight and natural mortality in juvenile and adult fish: A comparison of natural ecosystems and aquaculture. J. Fish Biol. 49: 627-647. Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis. 1997. A Pacific interdecadal oscillation with impacts on salmon production. Bull. Am. Meteorol. Soc. 78: 1069-1079. Marschall, E. A., and L. B. Crowder. 1995. Density-dependent survival as a function of size in juvenile salmonids in streams. Can. J. Fish. Aquat. Sci. 52:136-140. McCune, B„ and J. B. Grace. 2002. Analysis of ecological communities. 300 p. MjM Soft- ware Design, Gleneden Beach, OR. McCune, B., and M. J. Mefford. 1999. PC-ORD, multivariate analysis of ecological data, users guide, 237 p. MjM Software Design, Gleneden Beach, OR. McFarlane, G A., and R. J. Beamish. 2001. The re-occurrence of sardines off British Columbia characterises the dynamic nature of regimes. Prog. Oceanogr. 49:151-165. McGowan, J. A.. D. B. Chelton, and A. Conversi. 1999. Plankton patterns, climate, and change in the Cali- fornia Current. In Large marine ecosystems of the Pacific Rim: assessment, sustainability. and management (K. Sherman and Q. Tang, eds.), p. 63-105. Blackwell Sci- ence, Maiden, MA. McGurk, M. D. 1996. Allometry of marine mortality of Pacifc salmon. Fish. Bull. 94:77-88. Milner, G B., D. J. Teel, F. M. Utter, and G. A. Winans. 1985. A genetic method of stock identification in mixed populations of Pacific salmon, Oncorhynchus spp. Mar. Fish. Rev. 47:1-8. Myers, J. M., R. G. Kope. G J. Bryant, D. J. Teel, L. J. Lierheimer, T C. Wainwright, W S. Grant, F. W. Waknitz, K. Neely, S. T Lindley, and R. S. Waples. 1998. Status review of chinook salmon in Washington, Idaho, Oregon, and California. NOAA Tech. Memo. NMFS- NWFCS-35, 443 p. Nehlsen, W, J. E. Williams, and J. A. Lichatowich. 1991. Pacific salmon at the crossroads: stocks at risk from California. Oregon. Idaho, and Washington. Fisheries 16: 4-21. Nicholas, J. W., and D. G. Hankin. 1988. Chinook salmon populations in Oregon coastal river basins: Description of life histories and assessment of recent trends in run strengths. Oregon Department of Fish and Wildlife, Fish Division Information Report 88-1, 359 p. Paul. A. J., and T. M. Willette. 1997. Geographical variation in somatic energy content of migrating pink salmon fry from PWS: a tool to mea- sure nutritional status. In Proceedings of the interna- tional symposium on the role of forage fishes in marine ecosystems. Alaska Sea Grant College Program, Report 97-01, p. 707-720. Univ. Alaska, Fairbanks, AK. Pearcy. W. G 1992. Ocean ecology of North Pacific salmonids, 179 p. Univ. Washington Press, Seattle, WA. Pearcy, W G, R. D. Brodeur, and J. P. Fisher. 1990. Distribution and biology of juvenile cutthroat trout 46 Fishery Bulletin 102(1) (Oncorhynchus clarki clarki) and steelhead 10. mykiss) in coastal waters off Oregon and Washington. Fish. Bull. 88: 697-711. Pearcy, W. G.. and J. P. Fisher. 1988. Migrations of coho salmon, Oncorhynchus kisutch, during their first summer in the ocean. Fish. Bull. 86: 173-195. 1990. Distribution and abundance of juvenile salmonids off Oregon and Washington, 1981-1985. NOAA Tech. Rep. NMFS 93:1-83. Pella, J., and G. B. Milner. 1987. Use of genetic marks in stock composition analyses. In Population genetics and fishery management, (N. Ryman and F. Utter, eds.), p. 247-276. Washington Sea Grant, Univ. Washington Press, Seattle, WA. Perry, R. I.. B. Hargreaves, B. J. Waddell, and D. L. Mackas. 1996. Spatial variations in feeding and condition of juve- nile pink and chum salmon off Vancouver Island, British Columbia. Fish. Oceanogr. 5:73-88. Perry, R. I., and S. J. Smith. 1994. Identifying habitat associations of marine fishes using survey data: an application to the Northwest Atlantic. Can. J. Fish. Aquat. Sci. 51:589-602. Peterson, W. T., and J. E. Keister. 2002. The effect of a large cape on distribution patterns of coastal and oceanic copepods off Oregon and northern California during the 1998-1999 El Nino-La Nina. Prog. Oceanogr. 53:389-411. Post. J. R., and E. A. Parkinson. 2001. Energy allocation strategy in young fish: allometry and survival. Ecology 82:1040-1051. Ricker.W E. 1992. Back-calculation of fish lengths based on proportion- ality between scale and length increments. Can. J. Fish. Aquat. Sci. 49:1018-1026. Shenker, J. M. 1988. Oceanographic associations of neustonic larval and juvenile fishes and Dungeness crab megalopae off Oregon. Fish. Bull. 86:299-317. Sutton, S. G, T. P. Bult, and R. L. Haedrich. 2000. Relationship among fat weight, body weight, water weight, and condition factors in wild Atlantic salmon parr. Trans. Am. Fish. Soc. 129:527-538. Syrjala, S. E. 1996. A statistical test for a difference between the spatial distributions of two populations. Ecology 77:75-80. Teel, D. J., D. M. Van Doornik, D. R. Kuligowski, and W. S. Grant. 2003. Genetic analysis of juvenile coho salmon [Oncorhyn- chus kisutch) off Oregon and Washington reveals few Columbia River wild fish. Fish. Bull. 101:640-652. U.S. GLOBEC. 1994. A science plan for the California Current. U.S. GLOBEC Rep. 11. 134 p. Univ. California, Berkeley, CA. Weitkamp, L., and K. Neely. 2002. Coho salmon [Oncorhynchus kisutch I ocean migration patterns: insight from marine coded-wire tag recoveries. Can. J. Fish. Aquat. Sci. 59: 1 100- 1 1 1 5. Weitkamp, L. A., T. C. Wainright, G. J. Bryant, G. B. Milner, D. J. Teel, T G. Kope, and R. S. Waples. 1995. Status review of coho salmon from Washington. Oregon, and California. NOAA Tech. Memo. NMFS- NWFSC-24, 258 p. Appendix Table 1 Summary of releases of coho salmon smolts in 2000 by region. This summary of releases of all hatchery coho salmon smolts by region was calculated from data in the Pacific States Marine Fisheries Commission Regional Mark Information System (http://www.rmis.org/ [accessed 5 April 2003]) and in USFWS 2001 (see Footnote 2 in the general text). No. of release groups ToHl fish Release weight (gl released Marked mean I SD ) All British Columbia 250 13,612,715 71.4', 19.6(5.7) Washington: St. Juan de Fuca, Puget Sound, Skagit River, Nooksack River, etc. 83 15,316,299 86 r, 29.1 (19.7) Washington: North of Columbia River to Cape Flattery 63 7,630,257 76 7', 31.6(5.3) Columbia River 140 29,879,137 89.09i 32.0^ 1 Oregon Coast north of Cape Blanco 14 809,962 95.69! 41.6(7.41 Southern Oregon and Northern California: Rogue, Klamath, and Trinity Rivers 5 745.060 99.8^' 42.1 (4.4) ' 100% of the fish released from Klamath and Trinity Rivers were clipped on the maxillary. 47 Abstract— Between June 1995 and May 1996 seven rookeries in the Gulf of Cali- fornia were visited four times in order to collect scat samples for studying spa- tial and seasonal variability California sea lion prey. The rookeries studied were San Pedro Martir, San Esteban. El Rasito, Los Machos, Los Cantiles. Isla Granito, and Isla Lobos. The 1273 scat samples collected yielded 4995 otoliths (95.3%) and 247 (4.7%) cepha- lopod beaks. Fish were found in 97.4% of scat samples collected, cephalopods in 11.2%, and crustaceans in 12.7%. We identified 92 prey taxa to the species level, 11 to genus level, and 10 to family level, of which the most important were Pacific cutlassfish (Trichiuruslepturus), Pacific sardine (Sardinops caeruleus), plainfin midshipman (Porichthys spp. ), myctophid no. 1, northern anchovy (Engraulis mordax). Pacific mackerel (Scomber- japonicus), anchoveta (Ceten- graulis mysticetus), and jack mackerel (Trachurus symmetricus). Significant differences were found among rooker- ies in the occurrence of all main prey (P<0.04), except for myctophid no. 1 (P>0.05). Temporally, significant dif- ferences were found in the occurrence of Pacific cutlassfish, Pacific sardine, plainfin midshipman, northern an- chovy, and Pacific mackerel (P<0.05). but not in jack mackerel lx 2 =2.94, df=3, P=0.40 1, myctophid no. l(;r= 1.67, df= 3, P=0.64 ), or lanternfishes ( x 2 =2.08, df=3, P=0.56). Differences were observed in the diet and in trophic diversity among seasons and rookeries. More evident was the variation in diet in relation to availability of Pacific sardine. Spatial and temporal variation in the diet of the California sea lion (Zalophus californianus) in the Gulf of California, Mexico Francisco J. Garcia-Rodriguez David Aurioles-Gamboa Centra Interdisciplinary de Ciencias Mannas-lnstituto Politecnico Nacional Departamento de Biologia Manna y Pesquerias Apdo. Postal 592 La Paz, Ba|a California Sur, Mexico E-mail address (for F J. Garcia-Rodriguez) fjgrodriifflcibnor.mx Manuscript approved for publication 9 October 2003 by Scientific Editor. Manuscript received 20 October 2003 at NMFS Scientific Publications Office. Fish. Bull. 102:47-62 (2004). The population of the California sea lion (Zalophus californianus), in the Gulf of California numbers approxi- mately 23,000 individuals, 82% of which inhabit the northern region of the gulf above latitude 28° (Aurioles- Gamboa and Zavala-Gonzalez, 1994). In this region are found the most important reproductive areas and the highest pup production of the Gulf. Aurioles-Gamboa and Zavala-Gonzalez (1994) suggested that the high con- centration of animals in this region is related to high abundance of pelagic fish such as Pacific sardine (Sardinops caeruleus) (also known as South Ameri- can pilchard, FAO), Pacific mackerel (Scomber japonicus). Pacific thread herring (Opisthonema libertate), and anchoveta (Cetengraulis mysticetus) (Cisneros-Mata et al., 1987 1 ; Cisneros- Mata et al., 1991 2 ; Cisneros-Mata et al., 1997 3 ). Despite the importance of the north- ern gulf region, feeding studies of the California sea lion at Gulf of California rookeries have been few and have been conducted at different time periods. Researchers have studied sea lion diet in Los Islotes (Aurioles-Gamboa et al., 1984; Garcia-Rodriguez, 1995), Los Cantiles (Isla Angel de la Guarda), Isla Granito (Sanchez-Arias, 1992; Bautista- Vega, 2000), and Isla Racito (Orta-Davi- la, 1988). These studies have shown that sea lions consume a variety of prey and that differences exist between the diet of sea lions found at different rookeries within the Gulf of California. At Los Islotes, the most important prey were cusk eel (Aulopus bajacali), bigeye bass (Pronotogrammus eos), threadfin bass (Pronotogrammus multifasciatus), and splitail bass (Hemanthias sp.) (Aurioles- Gamboa et al, 1984; Garcia-Rodriguez. 1995). At Los Cantiles and Isla Granito important prey were lanternfish (Dia- phus sp.), northern anchovy (Engraulis mordax). Pacific cutlassfish (Trichiurus nitens), shoulderspot (Caelorinchus scaphopsis), and Pacific whiting (Mer- luccius productus) (Sanchez-Arias, 1992; Bautista-Vega, 2000), whereas at Isla Racito, important prey were Pacific sardine (Sardinops caeruleus). Pacific mackerel (Scomber japonicus), grunt (Haemulopsis spp.), rockfish (Sebastes 1 Cisneros-Mata, M. A.. J. P. Santos-Molina, J. A. DeAnda M.,A. Sanchez-Palafox, and J. J. Estrada. 1987. Pesqueria de sardina en el noroeste de Mexico ( 1985/86 ). Informe Tecnico, 79 p. Centro Regional de Inves- tigaciones Pesqueras de Guaymas. INP. SEPESCA. Calle 20 No. 605 Sur Col. La Cantera. Guaymas, Son. CP. 85400. 2 Cisneros-Mata, M. A., M. O. Nevarez- Martinez, G. Montemayor-Lopez, J. P. Santos-Molina, and R. Morales- Azpeitia. 1991. Pesqueria de sardina en el Golfo de California de 1988/89-1989/90. Informe Tecnico. 80 p. Centro Regional de Investigaciones Pesqueras de Guaymas. INP. SEPESCA. Calle 20 No. 605 Sur Col. La Cantera. Guaymas, Son. CP. 85400. 3 Cisneros-Mata, M. A., M. O. Nevarez- Martinez, M. A. Martinez-Zavala, M. L. Anguiano-Carranza, J. P. Santos-Molina, A. R. Godinez-Cota, and G. Montemayor- Lopez. 1997. Diagnosis de la pesqueria de pelagicos menores del Golfo de Califor- nia de 1991/92 a 1995/96. Informe Tecnico, 59 p. Centro Regional de Investigaciones Pesqueras de Guaymas. INP. SEMARNAP. Calle 20 No. 605 Sur Col. La Cantera. Guavmas, Son. CP. 85400. 48 Fishery Bulletin 102(1) spp. ), and Pacific whiting (Merluccius spp. ) (Orta-Davila, 1988). Some California sea lion prey are important fisheries resources in Mexico. The Pacific sar- dine, for example, is the target of a fishery be- gun in 1967 and which, along with the northern anchovy, contributed to most of the volume of the catch (200,870 t during the 1995-96 season) obtained in the Gulf (Cisneros-Mata et al. 3 ). The central and northern regions of the Gulf of California harbor the greatest abundance of sea lions and schooling fishes, such as the sar- dine and anchovy. Because of this, knowledge of sea lion feeding habits and their temporal and spatial variability is relevant to understanding the potential interaction between sea lions and fisheries in the area (Orta-Davila, 1988; San- chez-Arias, 1992; Bautista-Vega, 2000). In this article, we present the results of concurrent diet studies conducted at various rookeries and haulout areas of sea lions in the northern rookeries of the Gulf of California to examine the prey consumed, and spatial and temporal variability in their diet. Materials and methods 32° 28° 24° 20° 16° 12° Scat samples of California sea lions were collected at Isla San Pedro Martir (SPM, 28°24'00"N, 112°25'3"W), Isla San Esteban (EST, 28°42'00"N, 112°36'00"W), Isla Rasito (RAS, 28°49'30"N, 112°59'30"W), Isla Granito (GRA, 29°34'30"N, 113°32'15"W), Isla Lobos (LOB, 30°02'30"N, 114°. 28'30"W), and at two colonies of Isla Angel de la Guarda known as Los Machos (MAC, 29°20'00"N, 113°30'00"W), and Los Cantiles (CAN, 29°32'00"N, 113°29'00"W, Fig. 1). The total number of California sea lions in these seven rookeries was approximately 15,000 animals (that were hauled out) of which about 12.2% inhabit San Pedro Martir. 34.7% San Esteban, 2.8% El Rasito, 10.0% Los Machos, 8.7%. Los Cantiles, 11.0% Isla Granito, and 20.6% Isla Lobos (Aurioles-Gamboa and Zavala-Gonzalez, 1994). All the animals were spread out along the shoreline of each island, except at Isla Angel de la Guarda, where they were clustered within two areas: Los Cantiles, on the eastern shoreline and Los Machos on the western shoreline. Scat samples were obtained at reproductive and non- reproductive haulout areas between June 1995 and May 1996. At El Rasito, sampling was done only at one reproduc- tive area; fresh and dried samples were collected (Fig. 2). If for any reason a scat was not collected (because it was found in pieces or in poor condition), it was destroyed and the site was cleared to avoid collection during subsequent trips. All fresh and dried samples collected were pooled to represent each sampling period. We assumed that the diet information corresponded to a time period close to the col- lection trip, but some dried scats could have been deposited shortly after the last collection. Pacific Ocean 122° 118° 114° 110° 106° Figure 1 Map of Baja California showing location of California sea lion rook- eries that were studied in the Gulf of California. Scats were stored in plastic bottles and then dried shortly thereafter to prevent decomposition offish otoliths and other hard parts (which were used in subsequent prey identification) until the scats could be processed at a later date. The samples were processed by soaking in a weak biodegradable detergent solution for 1 to 7 days before being sifted through nested sieves of 2. 00-, 1.18-. and 0.5-mm mesh size. Fish bones and scales, eye lenses of fish and squid, otoliths, cephalopod beaks, and crustacean fragments were extracted from the samples. Cephalopod beaks were stored in 70% ethanol, and the other items were dried and stored in vials. Sagittal otoliths and cephalopod beaks were used to identify teleost fish and cephalopods, re- spectively. Identifications were made by using photographs and diagrams from Clarke (1962), Fitch ( 1966), Fitch and Brownell (1968), and Wolff (1984), as well as voucher specimen material from the 1) Center Interdiseiplinario de Marinas Ciencias (CICIMAR), 2) Instituto Tecnologico y de Estudios Superiores de Monterrey, Guaymas, 3) Los Angeles County Museum of Natural History, California, and 4) Centro de Investigacion Cientifica y de Educacion Superior de Ensenada (CICESE). Baja California, Mexico. Prey species identifed to family level were coded by using Garcia-Rodriquez and Aurioles-Gamboa: Spatial and temporal variation in the diet of Zalophus californianus 49 San Pedro Martir (SPM) 28° 24'- HA San Esteban (EST) 112°40' 112=38' 1 12=36' 112=34' 112=32' J L 28=44- 28=42' El Rasito (RAS) Angel de la Guarda 28=49' 113=40' 113=30' 113=20' 113=10' 113=00' I I i i i 29=30'- r^\ «— Los Cantiles (CAN) / (RAyHA) 29° 20'- A\ ^-. 29° 10- Los Machos (MAC)\ ^ (RA y HA) ^v \ M Isla Granito (GRA) Isla Lobos (LOB) 113' 34' 113° 33' i 29° 35'- s RA RA ha *\y 29=34'- 30=03'. 114=29' 1 114= 28 I | RA K HA - Figure 2 Location of sites where samples of California sea lion scats were collected at each island. RA = reproductive area; HA = haulout area. the family name plus a sequential number. Otoliths from prey species that were not identified to species, genus, or family level were coded with "fish species" plus a number. Three indices were used to describe the diet of sea lions. Percent number (PN) represents the percentage of the number of individuals for each prey taxon over the total number of individuals found in all scat samples. Percent of occurrence (PO) represents the percentage of scats hav- ing a given prey taxon and indicates the percentage of the population that is consuming a particular prey species. The third index, index of importance (IIMP) incorporates PN and PO and is defined as IIMP, 'T ^ u X (1) where x t = number of individuals of taxon z' in scatj; X = total number of individuals from all taxa found in scat J; and U = total number of samples with prey. The IIMP, developed for scat analysis (Garcfa-Rodriguez, 1999), was used to determine the importance of prey species, their spatial and temporal variation in the diet. 50 Fishery Bulletin 102(1) diversity of prey estimates, and measures of similarity among rookeries. Crustaceans were not incorporated into the IIMP index because it was not possible to quantify the number of individuals in the samples. We used the IIMP Index because it is less sensitive to changes in the number of prey in an individual scat com- pared to PN. For example, if a scat contains a single prey taxon, the IIMP does not change regardless of the number of individuals of that taxon, in that scat. However, as one increases the number of individuals of a given prey taxon in the scat, the PN will also increase for that prey. The IIMP allows each scat to contribute an equal amount of information, whereas PN can be dominated by a few scats with a large number of a single prey-taxon otoliths. In this manner the IIMP is similar to the split-sample frequency of ocurrence (SSFO) index, developed by Olesiuk (1993), where each scat is treated as a sampling unit and does not assume, as does PN, that the distribution of prey hard parts between scats is uniform. However, with the SSFO index, each prey taxon in a given scat is given an equal weight for that scat. If only one species occurs in a sample, its occurrence is scored as 1, if two species occur, each oc- currence is scored as 0.5, and so forth (Olesiuk, 1993). The IIMP index incorporates more information than the SSFO index, regardless of the number of individuals of each taxon in the scat. 4 Percent number (PN) was used only to show differences between broad prey groups (fishes and cephalopods) and PO was used to review the temporal and spatial changes from each main prey (those with average IIMP of at least 10% at any rookery). For all estimations, a single otolith (right or left) or single cephalopod beak (upper or lower) represented one individual prey. We tested the hypothesis that the occurrence of the main prey was independent of the rookery and of the date collection using contingency tables and an estimator of chi-square (x~) (Cortes, 1997). Total length of the otoliths (mm) and the linear equation obtained by Alvarado-Castillo 5 were used to estimate the length of the Pacific sardine (total length mm=7. 41+147. 24xotolith length mm); r=0.85, n=2740). Insufficient data did not allow estimating the size of speci- mens from May. All estimated lengths were transformed us- ing loglO, followed by an ANOVA among sampling periods. The size of Pacific sardine consumed by California sea lion was compared to those caught in the commercial fishery. We chose to estimate the size of Pacific sardines because of the broad information available concerning age and growth and other aspects about the fishery and because we found many sardine otoliths in good condition. Spatial and temporal correlation in composition of diet was compared by using the Spearman rank correlation co- 4 Garcfa-Rodriguez, F. J., and J. De la Cruz-Agiiero. In prep. An index to measure the specie prey importance of California sea lion ^Zalophus californianus) based on scat samples. 'Alvarado-Castillo, R. Unpubl. data. Departamento de Biologia y Pesquerias, Centro Interdisciplinary de Ciencias Marinas. Avenida IPN S/N Col. Palo Playa de Santa Rita, La Paz, Baja California Sur, Mexico 23070. efficient (R s ) (Fritz. 1974). Pairs of IIMP values were used for each prey taxon in a pair of sampling events (rookeries or sampling dates) to examine the correlation among them. This analysis was limited to prey that had an IIMP value of 10% or more. Cluster analysis of average IIMP data for the seven rookeries was used to assess the similarity of the diet among rookeries. The dendrogram for the cluster analysis was based on relative Euclidean distances and unweighted pair-grouping methods (UPGMA) strategy (Ludwig and Reynolds, 1988). We included only prey that, for at least one occasion, had IIMP values >10%. Trophic diversity was evaluated by using diversity curves (Hurtubia, 1973) developed from IIMP index data. For each date and colony, the cumulative diversity was calculated for increasing numbers of sequentially numbered (but we as- sumed randomly deposited and collected) scat samples. The diversity curves also allowed us to evaluate sample size (Hurtubia, 1973; Hoffman. 1978; Magurran, 1988, Cortes, 1997) by identifying the point at which the curve flattens. The diversity was estimated by using the Shannon index: H' -^P,\nPr (2) where H' = trophic diversity; S = total number of prey taxa; and P l = IIMP r and represents the relative abundance of taxon i obtained from each scat and pooled from scat 1 up to the total number of scats collected. The values of trophic diversity were then plotted against the number of pooled scats. Results Identification of prey The 1273 scat samples collected during June 1995 through May 1996 (Table 1) yielded fish remains in 97.4% of the samples, cephalopod remains in 11.2%, and crustacean remains in 12.7%. Fish and cephalopods represented 95.39; and 4.7%, respectively, of the 5242 individual prey (excluding crustaceans). The occurrence and number of these prey groups changed temporally and spatially (Fig. 3). We identified 92 prey taxa to the species level, 11 to the genus level, and 10 to family level from 851 scats (Table 2). Remaining scats had damaged prey structures in a condition that prevented us from identifying species to the genus or family level. We found nine main prey with IIMP average values a 10% (Table 3): the Pacific cutlassfish {THchiurus lepturus), the Pacific sardine (Sardinops eaeruleus), the plainfin mid- shipman (Porichthys spp.), myctophid no. 1, the northern anchovy iEngraulis mordax), Pacific mackerel {Scomber japonicus), the anchoveta (Cetengraulis mysticetus), jack mackerel iTrachurus symmetricus), and the lanternfish (unid. myctophid). Garcia-Rodriquez and Aurioles-Gamboa: Spatial and temporal variation in the diet of Zalophus californianus 51 Table 1 Number of scats collected at each rookery for each sampling period. June 1995 San Pedro Martir (SPM) SanEsteban(EST) ElRasito(RAS) Los Cantiles (CAN) IslaGranito(GRAl Los Machos (MAC) IslaLobos(LOB) Total 22 50 11 20 24 39 72 238 September 1995 January 1996 33 66 56 58 20 32 139 404 91 58 47 41 36 72 433 Mav 1996 29 67 25 35 19 23 198 Total 172 274 150 160 104 107 306 1273 Spatial and temporal variability of the main prey The importance (IIMP) of the Pacific cutlassfish was greater in Los Cantiles (28.4%), Isla Lobos (20.8%), and Isla Granito (48.5%) than at other sites (Fig. 4). The Pacific sardine was the dominant prey at Los Machos and to the south. There was a significant correlation across the sea- sons between Los Machos and El Rasito (r=0.998. P=0.04), but not between Los Machos and Isla Granito U-0.602, P=0.59). The IIMP of sardine was also correlated between El Rasito and San Esteban (r=0.95, P=0.04). The plainfin midshipman did not show a clear pattern, but its presence in the diet increased northeastward from Isla Angel de la Guarda. Lanternfishes, especially myctophid no. 1, were the dominant prey at San Pedro Martir, San Esteban, and El Rasito. The presence of Pacific mackerel was positively correlated with the presence of the Pacific sardine. The anchoveta was only found at Isla Lobos, and jack mackerel at El Rasito, San Pedro Martir, and Isla Granito. The changes in the PO of the main prey coincided with the variations of the IIMP. The occurrence of Pacific cut- lassfish. Pacific sardine, plainfin midshipman, northern anchovy, Pacific mackerel, and jack mackerel was signifi- cantly different (P<0.04) among rookeries. Myctophid no. 1 showed no significant difference in ocurrence 10% (Table 3) for a given collection. The Spearman rank correlation coefficient of IIMP between any pair of rookeries during June, September, January, and May was not significant (P>0.08). There was no positive correla- tion among any pair of sampling periods for any rookery (P>0.14), except between January and May at San Pedro Martir (P s =0.64, P<0.05) and El Rasito (P s =0.66, P<0.05) and between January and June as well as between Janu- ary and May at Isla Lobos (R s =0.56, P=0.05; and P s =0.59, P=0.05. respectively). The similarity in diet was related to the distance between rookeries. A clustering for the seven rookeries was obtained from these 25 prey taxa (Fig 6). We arbitrarily used a "cut- off" distance of 0.3 and 0.4 on the dendrogram as reference points for identifying clusters. The group obtained by us- ing the first distance (0.3) showed four feeding areas: one located in the south ( area I; San Pedro Martir, San Esteban, and El Rasito), another in Canal de Ballenas (area II: Los Machos) and two in the north (area III: Los Cantiles and Isla Lobos; and area IV: Isla Granito). When the second distance (0.4) was used, the seven rookeries grouped into two clusters: 1) the southern region and Canal de Ballenas, and 2) the region north of Angel de la Guarda. Spatial and temporal variability in trophic diversity Temporal variability in trophic diversity was evident between the rookeries (Fig. 7). In general, two patterns could be differentiated: one in which the diversity varied little throughout the year and the other in which diversity was high in January and low in September. The rookeries San Pedro Martir and Isla Lobos showed the first pattern and Los Machos and Isla Granito (and to a lesser extent, San Esteban and El Rasito) showed the second pattern. In September, when diversity was low, the dominant prey at 52 Fishery Bulletin 102(1) 100 80 60 40 20 100 T 80 60 40 20 0.- Percent number D Fishes Cephalopods JUNE 1995- MAY 1996 SPM EST RAS MAC CAN GRA LOB □ Fishes Cephalopods JUNE 1995 100 80 60 40 20 100 80 60 40 20 100 80 60 40 20 SPM EST RAS MAC CAN GRA LOB Fishes Cephalopods SEPTEMBER 1995 SPM EST RAS MAC CAN GRA LOB □ Fishes Cephalopods JANUARY 1996 SPM EST RAS MAC CAN GRA LOB □ Fishes Cephalopods MAY 1996 SPM EST RAS MAC CAN GRA LOB 100 80 1 60 40 1 20 100' 80 60 40' 20' 0. Percent occurrence Q Fishes Cephalopods □ Crustaceans JUNE 1995- MAY 1996 M XI Jl SPM EST RAS MAC CAN GRA LOB □ Fishes Cephalopods □ Crustaceans JUNE 1995 n n ^3*. SPM EST RAS MAC CAN GRA LOB □ Fishes Cephalopods D Crustaceans SEPTEMBER 1995 n Jl n 100, 80 60 1 40 1 20 SPM EST RAS MAC CAN GRA LOB O Fishes H Cephalopods D Crustaceans JANUARY 1996 SPM EST RAS MAC CAN GRA LOB D Fishes I Cephalopods D Crustaceans MAY 1996 n^Q SPM EST RAS MAC CAN GRA LOB Figure 3 Percent number (PNi and percent occurrence (POl index values for fishes, cephalopods, and crustaceans found in samples of California sea lion scats collected at seven rookeries in the Gulf of California, Mexico, for all sampling periods combined and for each sampling period. San Esteban, El Rasito, and Los Machos was Pacific sar- dine, whereas at Isla Granito, it was Pacific cutlassfish (Fig. 4 1. The curves obtained for Los Cantiles showed a clear pattern of diversity only in September, although the trend in the January curve would suggest a higher diver- sity in January than in September. Garcia-Rodriquez and Aurioles-Gamboa: Spatial and temporal variation in the diet of Zalophus califomianus 53 Table 2 Prey of California sea lion identified from scat samples collected at Isla San Pedro Martir, Isla San Esteban, Isla El Rasito, Los Cantiles, Isla Granito, Los Machos and Isla Lobos from June 1995 through May 1996. n ind. = = number of individuals in the sample; PN = percent number; n occurr = number of occurrences PO = percentage of occurrence; IIMP = index of importance. Scientific name Common name n Ind. PN n Occurr. PO IIMP Trichiurus lepturus Pacific cutlassfish 306 5.837 128 15.041 16.392 Sardinops caeruleus Pacific sardine 358 6.829 88 10.341 10.020 Porichthys spp. midshipman 456 8.699 95 11.163 9.297 Myctophid no. 1 lanternfish 714 13.621 119 13.984 7.901 Engraulis mordax northern anchovy 430 8.203 43 5.053 5.199 Scomber japonicus Pacific mackerel 103 1.965 42 4.935 3.403 Cetengraulis mysticetus anchoveta 410 7.821 15 1.763 2.404 Loliolopsis diomedeae squid 77 1.469 35 4.113 2.399 Trachurus symmetricus jack mackerel 111 2.118 41 4.818 2.273 Merluccius spp. Pacific whiting 55 1.049 25 2.938 2.206 Pontinus spp. scorpionfish 61 1.164 26 3.055 1.842 Enoploteuthid no. 1 squid 95 1.812 23 2.703 1.754 Caelorinchus scaphopsis shoulderspot 65 1.240 25 2.938 1.728 Octopus sp. no. 1 octopus 24 0.458 17 1.998 1.614 Sebastes macdonaldi Mexican rockfish 42 0.801 18 2.115 1.496 Citharichthys sp no. 1 sanddab 120 2.289 23 2.703 1.220 Fish species no. 1 — 49 0.935 25 2.938 1.153 Haemulopsis leuciscus white grunt 176 3.357 21 2.468 1.093 Peprilus snyderi salema butterfish 163 3.110 33 3.878 1.025 Prionotus spp. searobin 12 0.229 9 1.058 0.855 Prionotus stephanophrys lumptail searobin 53 1.011 14 1.645 0.814 Argentina sialis Pacific argentine 19 0.362 13 1.528 0.754 Fish species no. 2 — 55 1.049 27 3.173 0.737 Hemanthias peruanus splittail bass 60 1.145 22 2.585 0.602 Fish species no. 3 — 9 0.172 6 0.705 0.592 Micropogomas ectenes slender croaker 13 0.248 9 1.058 0.547 Lepophidium spp. cusk-eel 9 0.172 3 0.353 0.532 Fish species no. 4 — 10 0.191 3 0.353 0.511 Sebastes exsul buccanner rockfish 15 0.286 10 1.175 0.505 Cranchiid no. 2 Squid 20 0.382 12 1.410 0.501 Haemulon flaviguttatum yellowspotted grunt 11 0.210 3 0.353 0.468 Sela r cru men oph th aim us bigeye scad 24 0.458 12 1.410 0.431 Fish species no. 5 — 33 0.630 19 2.233 0.384 Paralabrax sp. no. 1 sea bass 9 0.172 5 0.588 0.373 Synodus sp. no. 3 lizardfish 10 0.191 3 0.353 0.341 Lepophidium prorates prowspine cusk-eel 5 0.095 4 0.470 0.335 Fish species no. 6 — 9 0.172 5 0.588 0.324 Synodus sp. no. 1 lizardfish 25 0.477 10 1.175 0.324 Octopus sp, no. 2 octopus 8 0.153 7 0.823 0.308 Gonatus berryi squid 5 0.095 5 0.588 0.274 Mugil cephalus striped mullet 1 0.019 1 0.118 0.265 Paranthias colonus Pacific creole-fish 1 0.019 1 0.118 0.265 Batistes polylepis finescale triggerfish 13 0.248 4 0.470 0.245 Physiculus nematopus charcoal mora 30 0.572 12 1.410 0.244 Hemanthias spp. sea bass 9 0.172 6 0.705 0.234 Fish species no. 7 — 10 0.191 8 0.940 0.233 Uroconger varidens conger eel 8 0.153 5 0.588 0.189 Larimus spp. drum 8 0.153 6 0.705 0.174 Apogon retrosella barspot cardinalfish 5 0.095 4 0.470 0.173 Dosidicus gigas squid 8 0.153 5 0.588 0.171 continued 54 Fishery Bulletin 102(1) Table 2 (continued) Scientific name Common name n Ind. PN n Occurr. PO IIMP Merluccius productus Pacific whiting 1 0.019 1 0.118 0.167 Fish species no. 8 — 2 0.038 2 0.235 0.159 Synodus sp. no. 2 lizardfish 12 0.229 5 0.588 0.132 Scorpaena sonorae Sonora scorpionfish 2 0.038 1 0.118 0.130 Eucinostomus spp. mojarra 13 0.248 5 0.588 0.129 Fish species no. 9 — 3 0.057 3 0.353 0.127 Cynoscion reticulatus striped weakfish 23 0.439 7 0.823 0.124 Fish species no. 10 — 10 0.191 1 0.118 0.122 Caulolatilus affinis bighead tilefish 4 0.076 3 0.353 0.114 Paralabrax auroguttatus goldspotted sand bass 18 0.343 4 0.470 0.110 Fish species no. 11 — 3 0.057 2 0.235 0.102 Cranchiid no. 5 squid 1 0.019 1 0.118 0.097 Bodianus diplotaenia mexican hogfish 1 0.019 1 0.118 0.087 Prionotus sp. no. 1 searonbin 2 0.038 2 0.235 0.087 Strongylura exilis California needlefish 1 0.019 1 0.118 0.083 Synodus spp. lizardfish 6 0114 5 0.588 0.146 Fish species no. 12 — 3 0.057 3 0.353 0.074 Fish species no. 13 — 2 0.038 1 0.118 0.065 Fish species no. 14 — 3 0.057 1 0.118 0.060 Fish species no. 15 — 2 0.038 1 0.118 0.058 Fish species no. 16 2 0.038 2 0.235 0.056 Porichthys sp. 1 midshipman 1 0.019 1 0.118 0.052 Fish species no. 17 — 5 0.095 3 0.353 0.049 Calamus brachysomus Pacific porgy 5 0.095 2 0.235 0.043 Fish species no. 18 — 1 0.019 1 0.118 0.042 Fish species no. 19 — 5 0.095 2 0.235 0.041 Ophididae no. 1 — 1 0.019 1 0.118 0.040 Fish species no. 20 — 5 0.095 3 0.353 0.039 Sebastes sinesis blackmouth rockfish 2 0.038 1 0.118 0.039 Symphurus spp. tonguefish 3 0.057 1 0.118 0.038 Fish species no. 21 — 2 0.038 1 0.118 0.036 Pronotogrammus multifasciatus threadfin bass 8 0.153 2 0.235 0.029 Fish species no. 22 — 2 0.038 2 0.235 0.027 Fish species no. 23 — 2 0.038 1 0.118 0.021 Orthopristis reddingi Bronze-striped grunt 16 0.305 1 0.118 0.020 Fish species no. 24 — 2 0.038 1 0.118 0.020 Fish species no. 25 — 1 0.019 1 0.118 0.016 Cranchiidae no. 4 squid 2 0.038 2 0.235 0.014 Fish species no. 26 — 2 0.038 2 0.235 0.014 Histioteuthis heteropsis squid 0.019 1 0.118 0.014 Scorpaenidae no. 1 — 0.019 1 0.118 0.011 Fish species no. 27 — 0.057 2 0.235 0.011 Fish species no. 28 — 0.019 1 0.118 0.010 Fish species no. 29 — 0.019 1 0.118 0.008 Cranchiidae no. 3 squid 0.019 1 0.118 0.006 Bollmannia spp. goby 0.019 1 0.118 0.006 Fish species no. 30 — 0.019 1 0.118 0.005 Cranchiidae no. 1 squid 0.019 1 0.118 0.004 Paralabrax maculatofasciatus spotted sand bass 0.019 1 0.118 0.003 Ophidian scrippsae basketweave cusk-eel 0.019 1 0.118 0.003 Physiculus spp. cod. codling, mora 2 0.038 1 0.118 0.003 Ophididae no. 2 — 4 0.076 1 0.118 0.002 Unid. Carangidae jacks 8 0.153 3 0.353 0.141 continued Garcia-Rodriquez and Aurioles-Gamboa: Spatial and temporal variation in the diet of Za/ophus californianus 55 Table 2 (continued) Scientific name Common name n Ind. PN n Occurr. PO IIMP Unid. Engraulidae anchovies 1 0.019 1 0.118 0.248 Unid. Haemulidae grunts 13 0.248 11 1.293 0.509 Unid. Labridae wrasses 1 0.019 1 0.118 0.005 Unid. Mycthophidae lanternifishes 216 4.121 71 8.343 4.895 Unid. Ophididae cusk-eel 2 0.038 1 0.118 0.098 Unid. Scianidae drums 13 0.248 9 1.058 0.643 Unid. Scorpaenidae scorpionfishes 30 0.572 18 2.115 1.078 Unid. Serranidae sea bass 13 0.248 6 0.705 0.176 Unid. Triglidae searobins 1 0.019 1 0.118 0.002 Unid. fishes 39 0.744 16 1.880 1.819 Unid. cephalopod 4 0.076 4 0.470 0.373 Unid. fishes (very eroded ) 381 7.268 231 27.145 Remains of cephalopods 14 1.645 Remains of crustaceans 162 19.036 Discussion Stomach acids attack otoliths, affecting their size and number and consequently the estimate of prey occurrence and importance. Erosion of otoliths during digestion has been analyzed in studies of pinnipeds in captivity. Bowen (2000) reviewed nine studies that estimated the propor- tion of otoliths recovered in scat samples to obtain a prey-number correction factor (NCF). He found that NCF is greater than 1.0 because many prey species are not recovered in the scat samples. Additionally, the erosion level can be significantly different among prey species (Bowen, 2000) because of differences in the shape and microstructure of otoliths. Therefore, estimates of biomass based on scat analysis should be carefully interpreted because the consumption of some prey species can be under- or overestimated. Correction factors are needed to compensate for differential erosion for the prey species of each pinniped. In this study the most important prey of California sea lions were pelagic fish with small, thin, and fragile otoliths (Nolf, 1993). The lanternfish also have small otoliths — perhaps smaller than those of any other prey taxa found in the scats. Their true importance in California sea lion feeding may be underestimated because of erosion caused by stomach acids (Da Silva and Neilson, 1985; Murie and Lavigne, 1985; Jobling and Breiby, 1986; Jobling, 1987; Toll- it et al., 1997). Similarly, the presence of cephalopods may have been underestimated because their jaws are composed of chitin, which is harder to digest, and frequently are re- gurgitated (Pitcher, 1980; Hawes, 1983). However, the high resistence to digestion of cephalopod beaks allows recovery of them in good shape. Thus they are a good choice to use in such diet analyses (Lowry and Carretta, 1999). A numerical index of prey species importance may over- or underestimate the dominance of prey species in the diet because it does not consider the weight of the prey. For IIMP, a numerical index that assumes a similar weight for all prey species, the true importance of the individual large prey in the diet can be underestimated and the importance of individual small prey can be overestimated. This prob- lem is also present when the PO, PN, and the SSFO index are used because these are all based only upon the number and occurrence of otoliths and cephalopods beaks. As when using PN. and the SSFO, the IIMP does not work for species that cannot be enumerated, such as crustaceans. Given the tendencies of the trophic diversity curves, the sample size was suitable in almost all cases. However, at San Pedro Martir a few more samples would have been useful to fully depict the diet. At Los Cantiles, except during September 1995, the samplings should have been more intense because the flattened portion of the diversity curves are not clear. The information, therefore, that comes from those samples could be biased. However, the number of scats that we analyzed contained a high percentage of the consumed species, especially the main prey. The results of this study indicate that the California sea lion consumed mainly fish and some crustaceans and cephalopods. According to the PN index, fish were more important than cephalopods in the diet of sea lions. In ad- dition, fish were more frequent (PO) than crustacean and cephalopods. Crustaceans were represented in a similar manner in scats from all rookeries. Cephalopods, however, were more important at San Pedro Martir and San Esteban, prob- ably because they are more common towards the southern gulf. Species of the suborder Oegopsida, which includes oceanic species (Roper and Young, 1975), were most com- monly found in scats from these rookeries. Orta-Davila (1988) and Sanchez-Arias (1992) have also noted the low consumption of cephalopods at the northern rookeries. Fish were the most diverse and commonly eaten prey. In contrast to cephalopods, fish were slightly less important in the southern region. The availability and abundance of the various prey resources influenced the diet of the sea lions in the Gulf 56 Fishery Bulletin 102(1) Table 3 Prey of California sea lions having IIMP index values alO^ in at leas t one sampling period for seven rookeries in the Gulf of Cali- fornia, Mexico Blank indicate that species were not recorded in diet; ' — " means unavailable data. Prey species June 1995 September 1995 January 1996 May 1996 Average San Pedro Engraulis mordax 29.7 2.1 0.5 8.1 Marti r myctophid no. 1 29.0 10.5 9.0 20.5 17.3 Porichthys spp. 11.2 2.0 6.8 15.5 8.9 Prionotus stephanophrys 0.6 3.3 3.3 10.9 4.5 enopleoteuthid no.l 27.3 0.8 7.0 Sebastes macdonaldi 10.4 2.6 Haeumulopsis leuciscus 16.7 6.0 5.7 San Esteban Trichiurus lepturus 24.9 3.4 3.0 7.8 Sardinops caeruleus 10.0 34.1 4.2 12.1 unid. Myctophidae 13.79 3.4 4.3 10.9 8.1 myctophid no. 1 2.8 11.8 8.9 18.8 10.6 enopleoteuthid no. 1 16.9 4.2 Sebastes macdonaldi 2.1 9.7 1.4 3.3 fish species no. 1 1.7 11.0 3.2 El Rasito Porichthys spp. 26.2 4.0 2.3 8.1 unid. Myctophidae 16.4 1.5 8.1 16.4 10.6 Scomber japonicus 13.8 3.2 3.7 2.5 5.8 Pontinus spp. 11.5 5.1 4.1 10.9 7.9 Octopus sp. no. 1 11.5 2.9 7.7 5.5 myctophid no. 1 6.6 5.1 21.4 6.8 10.0 Sardinops caeruleus 1.6 40.1 0.9 7.3 12.5 Trachurus symmetricus 22.0 5.0 23.4 12.6 Caelorinchus scaphopsis 3.6 13.5 10.5 6.9 Los Machos Sardinops caeruleus 21.0 64.1 16.8 — 34.0 Scomber japonicus 19.0 10.9 — 10.0 Merluccius spp. 15.4 8.2 — 7.9 Trichiurus lepturus 11.7 5.4 — 5.7 Sebastes macdonaldi 1.8 11.3 — 4.4 Los Cantiles Porichthys spp. 66.7 15.5 20.6 Trichiurus lepturus 22.2 38.2 53.1 28.4 Engraulis mordax 3.7 0.4 14.3 4.6 myctophid no. 1 17.6 4.8 5.6 Sardinops caeruleus 6.8 19.0 6.5 fish species no. 3 0.9 14.3 3.8 unid. fishes 0.9 19.0 5.0 unid. Scianidae 14.3 3.6 Lepophidium spp. 14. 3.5 Lo/iolopsis diomedcav 21.1 5.3 Isla Granito Engraulis mordax 49.3 7.8 14.3 Trichiurus lepturus 22.0 70.1 2.0 100.0 48.5 unid. myctophidae 1.7 1.1 12.6 3.9 Sardinops caeruleus 0.9 18.7 4.9 Porichthys spp. 0.5 18.2 4.6 5.8 Citharichthys sp. no. 1 21.7 5.4 Isla Lobos Cetengraulis mysticetus 32.7 0.1 6.8 27.8 16.9 Trichiurus lepturus 25.2 27.7 15.8 14.3 20.8 Porichthys spp. 9.0 10.3 23.2 35.5 19.5 Loliolopsis diomedeae 4.9 2.2 11.6 3.5 5.6 Peprilus snyderi 23.5 5.2 7.2 Garcia-Rodriquez and Aurioles-Gamboa: Spatial and temporal variation in the diet of Zalophus califomianus 57 100 80 60 SPM EST RAS MAC CAN GRA LOB 11111 Ml H l l i 1 ^_ ^B Trichiurus lepturus 20 « hJI L^l h^^^m. I a 5 fe'i a i fell a i fell Si ? fell & i fell & 5 fe'i & i | n 10 * SI 3 $ t SH to n Sl=5 to t Sin (0 t Sin to n 5 1 => to -> S 100 80 1 1 1 1 1 1 40 20 \_m i J - _« : _ i Sardinops caeruleus ■7 Q. ;r >- 1 Z Q_ Z >- 1 2 Q. z >- 1 z 0- Z >-'Z 0- Z >;! Z 0- Z >r'Z 0- z ^ 3 W * S't W 3 5 ' => co =5 5 ' =^ w> ^ 5 I= 3 « ^ 5 l= i W ^ 5' =5 w -> 5 100, 80 1 I 1 1 1 1 1 I 1 1 I 1 60 40 I ! - Porichthys spp 20 - | P- ' - ' — ^ iMI;ilil;iill;ili l;l ft i £;5 M 1;5 8l 1 100 80 60 40 20 Myctophidae no. 1 Z Cl Z >'Z 0- Z >"'z Q- Z V'z Q- Z i ' Z CL Z >' Z 0- Z b'z CL Z £ D 111 < (0 -» S -> CO ->5 CD C_> c a o Q. 1 CD CD ro c 100 80 60 40 20 Myctophidae z cl z >-'z n z v'zcl z v'z cl z >'z 0- z b'z cl z i'z cl z > i 8 1 1;1 8 1 i;l 8 1 S;i 8 1 I;3 8 1 3;3 8 1 I;r 8 1 i 100 80 60 Scomber japonicus o 20 __)■ Q_ z o. z v'z 0- z >-'z Q. z bz Q- z i'z Q. z >: ' z 0. z >: ' z Q- z >; TWn5TU)nS=50)-)S-iV)->S->WnS->W-)S->W-iS 100 80 60 40 20 ■_ Engraulis mordax Z 0. Z >-'z 0- Z >-'z D- z >-'z 0. Z i'z Q- z £ ' Z 0- Z £: Z 0. Z >: t CO n S =) CO * S ' =j 0) n So CO n S =3 CO n S =5 CO n S =i CO =3 5 100 80 40 20 -m Cetengraulis mysticetus i & i fell a s s!i a ? si? & s fell a i fell a i fell a ? | T CO n 5 n CO * S n CO n St » n S n (0 n St 0) n 5 ( => to -, S 100 80 40 20 ^_ _L ^_ _L _^ -L Trachurus symmetricus Z 0- Z >".Z tL Z >,Z D- Z >-.Z 0- Z >ZtZ 0- Z ^,Z 0- Z £,Z Q- Z 5j D S tf -t'D QJ < D Ul < to * S|=i CO t S,=5 to n S,n to t S,=i to n S r n W n S spm : est : ras : mac : can : GRA : LOB Figure 4 Index of importance (IIMP) for nine prey species identified from samples of California at seven rookeries in the Gulf of California, Mexico, during June and September 1995 sea lions scats collected . and January and May 1996. of California. The distribution pattern of Pacific sardine closely agrees with its importance in the sea lions diet. The Pacific sardine occurred in high concentrations around Angel de la Guarda and Isla Tiburon during the summer and along the coast of southern Sonora during the winter, where spawning occurs (Cisneros-Mata et al. 3 ). Sardines 58 Fishery Bulletin 102(1] were consumed in the Canal de Ballenas region during the summer (September), when they are very abundant. Larger size Pacific sardines were consumed by sea lions most frequently during the summer when adult sardines occur more frequently in the Canal de Ballenas. As adult sardine left Canal de Ballenas ( Cisneros-Mata et al., 1997 ), the proportion of young individuals in the diet of sea lions increased. The fish eaten by sea lions were apparently smaller than those captured by the commercial fisher- ies. The average estimated size of the sardines consumed was 150.4 mm, whereas the average size of commercially caught fish during the 1995-96 season was 162.4 mm (Cis- neros-Mata et al. 3 ). This 7% difference in size may have been caused by an underestimation of otolith size because of digestive erosion ( Jobling and Breiby, 1986). If this is so, then the size of Pacific sardines consumed by sea lions is similar to the size of those captured by the fishery. Isla Lobos was the only rookery where Pacific sardine was not consumed. This finding differs from those of Cisneros- Mata et al. 3 which show the Pacific sardines present as far north as Isla Lobos. However, their study period was during the 1991-92 El Nino episode, whereas our study occurred during normal oceanographic conditios in 1995-96. Less is known about the spatial and temporal availability of other important prey. As with commercial captures (Arvizu-Martinez, 1987), Pacific mackerel occurred together with Pacific sardine. Similar varia- tions in occurrence for both species have been noticed from stomach content analyses of the giant squid (Dosidicus gigas) (Ehrhardt, 1991). Lanternfishes were abundant north of Isla Angel de la Guarda (Robison, 1972); however they were not im- portant in the diet of the California sea lion in this region. Their greater importance in the diet at southern rookeries was probably due to the absence of more preferred prey such as Pacific sardine, Pacific cutlass- fish, or anchoveta. The consump- tion of northern anchovy tended to be less important towards Canal de Ballenas, where Pacific sardine reached its maximum importance. The low spatial overlap of these two species has also been noted in other studies. The anchoveta was present only at Isla Lobos. This is an estuarine-lagoon species, typical of coastal lagoons of northern Sinaloa and Sonora (Castro-Aguirre et al., 1995). The presence of this prey in Isla Lobos is possibly due to the sandy coast (Walker, 1960), which is similar to that of the Sinaloa-Sonora coast. The diet of California sea lions differed among rooker- ies, probably due to differences in feeding sites and prey availability. Antonelis et al. (1990) studied the foraging characteristics of the northern fur seal (Callorhinus ur- sinus) and the California sea lion at San Miguel Island and found differences between foraging areas among 0.15 200-i 180- 160- 140- | 120- •£_ 100- £ 80- _J 60- 40- 20- n=121 JUN95 SEP95 JAN95 Figure 5 Size of Pacific sardine iSardinops caeruleus) estimated from otoliths found in California sea lions scats collected in Isla San Esteban, El Rasito, Granito, Los Cantiles, and Los Machos. One standard deviation is indicated from each mean. 0.2 0.25 0.3 0.35 0.4 0.45 Figure 6 Dendrogram of cluster analysis of seven rookeries determined with Euclidean dis- tance (computed from the IIMP of the 25 prey that had on at least one occasion a value >10%) and the UPGMA (unweighted pair-grouping methods) strategy. The vertical lines represent the points of references to delimit the groups. species. The northern fur seal was found most frequently foraging in oceanic water within 72.4 km from the island, whereas Califorinia sea lions forgaged more often in the shallower neritic zone, within 54.2 km from the island. Different foraging distances in California sea lions from San Miguel Island were found by Melin and DeLong ( 1999). During the nonbreeding season a higher percent- age of foraging locations occurred at distances less than 100 km, whereas during the breeding season most of the foraging locations occurred at distances greater than 100 km. These differences are probably due to the in- creased California sea lion population in San Miguel; this increase in population forces sea lions to exploit new areas as a density-dependent response to population Garcia-Rodriquez and Aurioles-Gamboa: Spatial and temporal variation in the diet of Zalophus californianus 59 SPM ■Jun95 »Sep96 ■4 I I I I I I I I I I I I I I I I I I I I I I t I 2 4 6 8 10 G M 16 18 2022242628303234 36 384042'! EST 2 4 6 8 10 12 2022 24 26283032 34 36 384042444648 50 RAS 2 4 6 8 10 12 14 16 18 202224262830323436384042444648 50 3 50 MAC 3 00 . * " * 2 50 . t „ / 2 00 . 1 50 . */ i - - - Jan t 00 . 50 . 1 t 1 1 1 1 1 1 1 1 * I I l l l l I I l ■t-f H- ■h-i 2 A 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 Sample size CAN I'l I I I i I i i i I I I I I I I I I i I I I I l l 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 GRA I ' l i i i i i i i i i i i i i i i i i i i 0246610121416 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 3 50 - LOB 3 00 . 2 50 . _, _. - -J—v 2 00 . r v J C^y 1 50 . 100 . . A /• - - - Jan96 0.50 •_i/ v - - May9 /, i -t-t- Mill i i i i i i i 0246810121416 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 40 50 Sample size Figure 7 Trophic diversity curves for California sea lions determined from scat samples collected at seven rookeries in the Gulf of California, Mexico. SPM = San Pedro Martir; EST = San Esteban; RAS = Isla Rasito; MAC = Los Machos; CAN = Los Cantiles; GRA = Isla Granito; LOB = Isla Lobos. 60 Fishery Bulletin 102(1) growth. Although, these differences could also be due to variability in the distribution of prey (Melin and DeLong, 1999), as suggested by Antonelis and Fiscus (1980), forag- ing areas might change with season and annual variations in prey availability and abundance. Foraging areas in the Gulf of California could lie closer to rookeries than those recorded for San Miguel Island sea lions because the diet was different among rookeries in spite of the shorter distance between them (54.2 km). At Los Islotes, Baja California Sur, adult females fed within 20 km of the colony (Duran-Lizarraga. 1998). Kooyman and Trillmich (1986a, 1986b) reported similar data in sea lion colonies of the Galapagos Islands. In the northern region of the Gulf of California, feeding range could be shorter than that at Los Islotes because of the higher concentration of food at high nutrient concentrations (phosphate, nitrate, nitrite, and silicate) in Canal de Ballenas that is associated with strong tidal mixing (Alvarez-Borrego, 1983). Four foraging zones were discerned from dietary differ- ences in sea lions from the seven rookeries studied. Zone I, which included San Pedro Martir. San Esteban. and El Rasito, was characterized by the consumption of lantern- fish; zone II, which included Los Machos was characterized by the consumption of Pacific sardine and Pacific mackerel; zone III, which included Isla Granito, by the consumption of Pacific cutlassfish and the northern anchovy; and zone IV, Los Cantiles and Isla Lobos, was characterized by the consumption of the plainfin midshipman and the Pacific cutlassfish. These four zones may indicate differences in habits used by sea lions or may indicate different oceano- graphic conditions exploited by sea lions. The eastern coast of the Gulf of California displays high photosyn- thetic pigment concentrations, associated with upwelling induced by winds from the northwest in the winter. These conditions may make Canal de Ballenas one of the most important for the distribution of Pacific sardine during the summer. Trophic diversity varied spatially and temporally. San Pedro Martir and Isla lobos sea lions seem to depend on a more stable feeding areas compared to sea lions at rook- eries on Isla Granito and Los Machos, where changes in diversity of consumed species indicated that sea lions feed on fewer species during certain times of the year. Similar results in relation to the changes in diversity were also noticed in the rookeries of the Channel Islands and Faral- lon Islands, California (Bailey and Ainley, 1982; Antonelis et al., 1984; Lowry et al., 1990; Lowry et al., 1991 ). Perhaps the tendency to have the highest values of diversity and little seasonal variation at San Pedro Martir is the result of this rookery being located in a zone of transition between two biogeographical areas. This geographical position con- fers greater environmental heterogeneity and greater ecological diversity (Walker, 1960). California sea lions in the upper region of the Gulf of California obtain the main portion of their diet from a relatively small number of species. The decrease in abun- dance of any of these food resources can seriously affect the population, particularly at Isla Granito and Los Machos because sea lions from these rookeries depend on a few species. Acknowledgments We wish to thank Secretaria de Marina, Armada de Mexico, for its great support during the field activities, and the Consejo Nacional de Ciencia y Tecnologia (CONACYT) for funding this study under grant number 26430-N. The Secretaria de Medio Ambiente, Recursos Naturales y Pesca (SEMARNAP) provided permits for field work (DOO.-700- (2)01104 and DOO.-700(2).-1917). We would like to thank Robert Lavenberg and Jeff Siegel for allowing us the use of otoliths from the collection at the Natural Museum His- tory of Los Angeles County and also Lawrence Barnes for his logistical support during the stay of first author at Los Angeles; we also thank Manuel Nava for allowing us the use of otoliths from the collection in Tecnologico de Monterrey, Campus Guaymas. We are also grateful to Unai Markaida for his assistance in prey identification based on the examination of cephalopods beaks. We thank Mark Lowry for commenting on an earlier draft of the paper, Norman Silverberg for reviewing the manuscript in English, and two anonymous reviewers for their valuable suggestions and criticism. The first author would like to thank Centro Interdisciplinario de Ciencias Marinas-IPN for a scholar- ship (PIFI, Programa Institucional para la Formacion de Investigadores) assigned for postgraduate studies. Literature cited Alvarez-Borrego, S. 1983. Gulf of California. In Estuaries and enclosed seas (B. H. Ketchum. ed.i, p 427-449. Elsevier Scientific Publish- ing Co.. Amsterdam. Antonelis, G. A, and C. H. Fiscus. 1980. The pinnipeds of the California Current. Calif. Coop. Oceanic Fish. Invest. Rep. 21:68-78. Antonelis, G A.. C. H. Fiscus. and R. L. DeLong. 1984. Spring and summer prey of California sea lions. Zalophus californianus, at San Miguel Island, California 1978-79. Fish. Bull. 82:67-75. Antonelis, G. A., B. S. Stewart, and W. F. Perryman. 1990. Foraging characteristics of females northern fur seals (Callorhinus ursinus) and California sea lion \Zalophus californianus). Can J. Zool. 68:150-158. Arvizu-Martinez, J. 1987. Fisheries activities in the Gulf of California. Mexico. Calif. Coop. Oceanic Fish. Invest. Rep. 28:32-36. Aurioles-Gamboa, D., C. Fox, F. Sinsel, and G. Tanos. 1984. Prey of sea lions iZalophus californianus) in the bay of La Paz, Baja California Sur. Mexico. J. Mammal. 65(3): 519-21. Aurioles-Gamboa, D., and A. Zavala-Gonzalez. 1994. Algunos factores ecolbgicos que determinan la distri- bucion y abundancia del lobo marino de California, Zalo- phus californianus. en el Golfo de California. Ciencias Marinas 20(4):535-553. Bailey, K. M., and D. G. Ainley. 1982. The dynamics of California sea lion predation on Pacific hake. Fish. Res. 1:163-176. Bautista-Vega, A. A. 2000. Vartiacion estacional en la dieta del lobo marino comun. Zalophus californianus. en las Islas Angel de la Guarda y Granito. Golfo de California, Mexico. Tesis de Licenciatura. Garcfa-Rodrfquez and Aurioles-Gamboa: Spatial and temporal variation in the diet of Zalophus californianus 61 1 13 p. Universidad Nacional Autonoma de Mexico. Mexico, D.F. Bowen, W. D. 2000. Reconstruction of pinniped diets: accounting for com- plete digestion of otoliths and cephalopod beaks. Can. J. Fish. Aquat. Sci. 57:898-905. Castro-Aguirre. J. L„ E. F. Balart, and J. Arvizu-Martinez. 1995. Contribucibn al conocimiento del origen y distribution de la ictiofauna del Golfo de California, Mexico. Hidro- biolbgica 5(1-21:57-78. Clarke, M. R. 1962. The identification of cephalopod "beaks" and the rela- tionship between beak size and total body weight. Bull. British Museum (Natural History) Zoology 8( 101:419-480. Cortes, E. 1997. A critical review of methods of studying fish feeding based on analysis of stomach contents: application to elas- mobranch fishes. Can. J. Fish. Aquat. Sci. 54:726-738. Da Silva, J., and J. D. Neilson. 1985. Limitations of using otoliths recovered in scats to estimate prey consumptions in seals. Can. J. Fish. Aquat. Sci. 42:1439-1442. Duran-Lizarraga. M. E. 1998. Caracterizacibn de los patrones de buceos de alimen- tation de lobo marino Zalophus californianus y su relacibn con variables ambientales en la Bahia de La Paz, B.C.S. Tesis de Maestria, 82 p. CICIMAR, La Paz, B.C.S., Mexico. Ehrhardt, N. M. 1991. Potential impact of a seasonal migratory jumbo squid (Dosidicus gigas) stock on a Gulf of California sardine (Sardinops sagax caerulea) population. Bull. Mar. Sci. 49(1-21:325-332. Fitch, J. E. 1966. Additional fish remains, mostly otoliths, from a Pleis- tocene deposit at Playa del Rey, California. Los Angeles County Mus., Cont. in Sci. (1191:1-16. Fitch, J. E., and R. L. Brownell. 1968. Fish otoliths in Cetacean stomachs and their impor- tance in interpreting feeding habits. J. Fish. Res. Board Canada. 25(121:2561-2574. Fritz, E. S. 1974. Total diet comparison in fishes by Spearman rank cor- relation coefficients. Copeia (19741:210-214. Garcia-Rodriguez, F. J. 1995. Ecologia alimentaria del lobo marino de California, Zalophus californianus californianus en Los Islotes, B.C.S., Mexico. Tesis de Licenciatura, 106 p. Univer- sidad Autonoma de Baja California Sur. La Paz, B.C.S. , Mexico. 1999. Cambios espaciales y estacionales en la estructura tro- fica del lobo marino de California, Zalophus caliornianus, en la region de la grande islas, Golfo de California. Tesis de Maestria, 73 p. CICIMAR. La Paz, B.C.S, Mexico. Hawes, D. 1983. An evaluation of California sea lion scat samples as indicators of prey importance. Master's thesis, 50 p. San Francisco State Univ., San Francisco, CA. Hoffman, M. 1978. The use of Pielou's method to determine sample size in food studies. In Fish food habits studies: proceedings of the second Pacific Northwest technical workshop (S. J. Lipovsky and C. A. Simenstad, eds.), p 56-61. Washington Sea Grant Publication, Seattle, WA. Hurtubia, J. 1973. Trophic diversity measurement in sympatric preda- tory species. Ecology 54(41:885-890. Jobling, M. 1987. Marine mammal faeces samples as indicators of prey importance — a source of error in bioenergetics studies. Sarsia 72:255-260. Jobling, M., and A. Breiby. 1986. The use and abuse of fish otoliths in studies of feeding habits of marine piscivores. Sarsia 71:265-274. Kooyman, G L., and F. Trillmich. 1986a. Diving behavior of Galapagos fur seals. In Fur seals: maternal strategies on land and at sea (R. L. Gentry and G. L. Kooyman, eds), p. 186-195. Princeton Univ. Press, New Jersey, NJ. 1986b. Diving behavior of Galapagos sea lions. In Fur seals — maternal strategies on land and at sea (R. L. Gentry and G. L. Kooyman, eds. 1, p. 209-220. Princeton LIniv. Press, Princeton. NJ. Lowry, M. S., and J. V. Carretta. 1999. Market squid (Loligo opalescens) in the diet of California sea lions (Zalophus californianus) in southern California ( 1981-1995). Calif. Coop. Oceanic Fish. Invest. Rep. 40:196-207. Lowry, M. S„ C. W. Oliver, C. Macky, and J. B. Wexler. 1990. Food habits of California sea lions Zalophus califor- nianus at San Clemente Island, California, 1981-86. Fish. Bull. 88:509-521. Lowry, M. S., B. S. Stewart, C. B. Heath, P. K. Yochem. and J. M. Francis. 1991. Seasonal and annual variability in the diet of Cali- fornia sea lions Zalophus californianus at San Nicolas Island, California, 1981-86. Fish. Bull. 89:331-336. Ludwig, J. A., and J. F Reynolds. 1988. Statistical ecology: a primer on methods and comput- ing. 338 p. John Wiley and Sons, New York, NY. Magurran, A. E. 1988. Ecological diversity and its measurement, 178 p. Princeton Univ. Press, Princeton, NJ. Melin, S. R., and R. L. DeLong. 1999. At-sea distribution and diving behavior of California sea lion females from San Miguel Island, California. In Proceeding of the fifth California islands symposium (D. R. Browne, K. L. Mitchell, and H. W. Chaney, eds.), p. 407-402. Santa Barbara Museum of Natural History, Santa Barbara, CA. Murie, D. J., and D. M. Lavigne. 1985. A technique to recovery of otoliths from stomach contents of piscivorous pinnipeds. J. Wildl. Manag. 49: 910-912. Nolf, D. 1993. A survey of perciform otoliths and their interest for phylogenetic analysis, with an iconographic synopsis of the Percoidei. Bull. Mar. Sci. 52(1 ):220-239. Olesiuk, P. F. 1993. Annual prey consumption by harbor seals (Phoca vitulina ) in the Strait of Georgia, British Columbia. Fish. Bull. 91:491-515. Orta-Davila, F. 1988. Habitos alimentarios y censos globales del lobo marino (Zalophus californianus) en el Islote El Racito, Bahia de las Animas, Baja California. Mexico durante octubre 1986-1987. Tesis de Licenciatura, 59 p. Universidad Autonoma de Baja California. Ensenada, B.C. Pitcher, K. W. 1980. Stomach contents and feces as indicators of harbour seal, Phoca vitulina, foods in the Gulf of Alaska. Fish. Bull. 78:797-798. 62 Fishery Bulletin 102(1 Robison. B. H. 1972. Distribution of the midwater fishes of the Gulf of California. Copeia (19721:449-61. Roper. C. F. E., and R. E. Young. 1975. Vertical distribution of pelagic cephalopods. Smith- sonian Contribution to Zoology 209(51 1:31. Sanchez-Arias, M. 1992. Contribucion al conocimiento de los habitos alimen- tarios del lobo marino de California Zalophus califomianus en las Islas Angel de la Guarda y Granito, Golfo de Cali- fornia, Mexico. Tesis de Licenciatura, 63 p. Universidad Nacional Autonoma de Mexico. Mexico, D.F. Tollit, D. J., M. J. Steward, P. M. Thompson. G. J. Pierce, M. B. Santos, and S. Hughes. 1997. Species and size differences in the digestion of oto- liths and beaks: implications for estimates of pinniped diet composition. Can. J. Fish. Aquat. Sci. 54:105-119. Walker, B. W. 1960. The distribution and affinities of the marine fish fauna of the Gulf of California. System. Zool. 9(3):123-133. Wolff, G A. 1984. Identification and estimation of size from the beaks of 18 species of cephalopods from the Pacific Ocean. NOAA Tech. Rep. NMFS 17, 49 p. 63 Abstract— Recruitment of bay anchovy {Anchoa mitchilli) in Chesapeake is related to variability in hydrologi- cal conditions and to abundance and spatial distribution of spawning stock biomass (SSB I. Midwater-trawl surveys conducted for six years, over the entire 320-km length of the bay, provided information on anchovy SSB, annual spatial patterns of recruitment, and their relationships to variability in the estuarine environment. SSB of anchovy varied sixfold in 1995-2000; it alone explained little variability in young-of-the-year (YOY) recruitment level in October, which varied ninefold. Recruitments were low in 1995 and 1996 (47 and 31xl0 9 ) but higher in 1997-2000 (100 to 265 xlO 9 ). During the recruitment process the YOY popu- lation migrated upbay before a subse- quent fall-winter downbay migration. The extent of the downbay migration by maturing recruits was greatest in years of high freshwater input to the bay. Mean dissolved oxygen (DO) was more important than freshwater input in controlling distribution of SSB and shifts in SSB location between April- May (prespawningl and June-August (spawning) periods. Recruitments of bay anchovy were higher when mean DO was lowest in the downbay region during the spawning season. It is hypothesized that anchovy recruit- ment level is inversely related to mean DO concentration because low DO is associated with high plankton produc- tivity in Chesapeake Bay. Additionally, low DO conditions may confine most bay anchovy spawners to the downbay region, where production of larvae and juveniles is enhanced. A modified Ricker stock-recruitment model indicated den- sity-compensatory recruitment with respect to SSB and demonstrated the importance of spring-summer DO levels and spatial distribution of SSB as con- trollers of bay anchovy recruitment. Recruitment and spawning-stock biomass distribution of bay anchovy (Anchoa mitchilli) in Chesapeake Bay* Sukgeun Jung Edward D. Houde University of Maryland Center for Environmental Science Chesapeake Biological Laboratory 1 Williams St., P.O. Box 38 Solomons, Maryland 20688 E-mail address (for S Jung): iung@cbl.umces.edu Manuscript approved for publication 30 September 2003 by Scientific Editor. Manuscript received 20 October 2003 at NMFS Scientific Publications Office. Fish. Bull. 102:63-77 (20041. Recruitment for marine fishes is vari- able and is regulated or controlled by a combination of density-dependent and density-independent processes. It has been hypothesized that density-inde- pendent processes dominate from the egg to larval stages whereas density- dependent control by predation may be more important in the juvenile stage (Sissenwine, 1984; Houde, 1987). Den- sity-dependent processes may be stock dependent, regulated by adult abun- dances, or dependent on abundances of the early-life stages (Ricker, 1975). In estuarine systems, where hydrological conditions (e.g. dissolved oxygen, tem- perature, and circulation) vary widely, the roles of density-independent physi- cal factors on fish recruitments may be dominant, making it difficult, but still important, to partition density- dependent and density-independent processes, particularly for short-lived small pelagic fishes such as anchovies and sardines. Bay anchovy {Anchoa mitchilli) (En- graulidae) is a coastal species distrib- uted broadly in the western Atlantic from Maine to Mexico. This small fish is the most abundant and ubiquitous fish in Chesapeake Bay, the largest estu- ary on the east coast of North America (Houde and Zastrow, 1991; Able and Fahay, 1998). It is not fished, yet there is evidence that recruitment is variable (Newberger and Houde, 1995). It feeds on zooplankton — primarily copepods and other small Crustacea — and is a major prey of piscivores, including several eco- nomically important fishes (Baird and Ulanowicz, 1989; Luo and Brandt, 1993; Hartman and Brandt, 1995). Male and female bay anchovy in Chesapeake Bay mature at 40^15 mm fork length (44-50 mm total length) at about 10 months of age, and peak spawning occurs in July (Zastrow et al.. 1991). Most eggs are produced by age-1 individuals (Luo and Musick, 1991; Zastrow et al., 1991). Bay anchovy may survive to age 3+ and reach approximately 100 mm length and 5 g wet weight ( Newberger and Houde, 1995; Wang and Houde, 1995). Newberger and Houde (1995) noted large differences in annual survey abundances of bay anchovy that appar- ently resulted from variability in an- nual recruitments. In Chesapeake Bay, abundance, growth, and mortality rates of bay anchovy eggs and larvae vary temporally and spatially (Dorsey et al, 1996; MacGregor and Houde, 1996; Rilling and Houde, 1999a, 1999b). Indi- vidual-based models were developed to test the hypothesis that recruitment of bay anchovy is determined by variable growth and mortality during early-life stages that are regulated by density-de- pendent processes (Wang et al., 1997; Cowan et al., 1999; Rose et al„ 1999). In previous research, there was little knowledge of levels of spawning stock biomass or density-independent envi- ronmental factors that may control re- cruitment through their effects on spa- tial and temporal variability in growth and mortality of prerecruit anchovy. * Contribution 3696 of the University of Maryland Center for Environmental Sci- ence, Chesapeake Biological Laboratory, Solomons, MD 20688. 64 Fishery Bulletin 102(1) 39°N 38°N ^vt^w N 37°N gi quehanna Upper — Middle Lower Atlantic Ocean 77°W 76°W Figure 1 Chesapeake Bay and stations sampled by the midwater trawl in the 1995-2000 surveys. Horizontal lines indicate boundaries of three designated regions. We evaluated environmental factors, spatial distribution of spawning stock biomass (SSB), and possible ontogenetic migrations of prerecruits (Dovel, 1971; Loos and Perry. 1991; Wang and Houde, 1995; Kimura et al, 2000) with respect to bay anchovy recruitment variability. Our objec- tives were 1) to estimate annual and regional variability in bay anchovy recruitment. 2) to evaluate effects of hy- drological conditions (mainly, freshwater input, and dis- solved oxygen concentration) on stage-specific distribution, ontogenetic migration, and recruitment, and 3) to identify mechanisms and describe patterns or trends in the bay anchovy recruitment process. Data were obtained in a six-year, multidisciplinary research program conducted throughout Chesapeake Bay. Materials and methods Study area Chesapeake Bay is a coastal plain estuary of partially mixed fresh water and sea water. Its 320-km mainstem varies in width from about 6 to 50 km (Fig. 1 ). The Bay is shallow; less than 10' r of its area is >18 m deep and approximately 50' i is <6 m deep. More than 809& of the freshwater entering the bay is from tributaries on its northern and western sides (Chesapeake Bay Program 1 ). Salinity grades from near-full seawater at the mouth of the bay to freshwater near its head. Water temperatures reach 28-30°C in mid summer, and fall to 1^°C in late winter (Murdy et al, 1997 ). Despite shallow depth, the bay usually has a strongly developed pycnocline, and has seasonally strong vertical gradients in temperature, salinity, and dissolved oxygen. Surveys Trawl surveys were conducted three times annually over the entire bay (April-May, June-August, and October). 1995-2000 (Table l.Fig. 1). Midwater-trawl (MWT) fish col- lections 2 were made on transects in three regions: the lower bay (37°05'N-37°55'N), middle bay (37°55'N-38°45'N I, and upper bay (38°45'N-39°25'N). As defined, the lower bay contains 51% of water volume, the middle bay 32^ .and the upper bay 17^ (Fig. 1). The number of midwater trawl sta- Chesapeake Bay Program. 2000. Chesapeake Bay: Introduc- tion to an ecosystem. U.S. Environmental Protection Agency, publ. EPA 903-R-00-001. 30 p. EPA. 410 Severn Ave, Suite 109, Annapolis. MD 21403. Trophic interactions in estuarine systems, midwater trawl sur- vey. University of Maryland Center for Environmental Sci- ence, Chesapeake Biological Laboratory, http://www.ch.esa peake.org/ ties/mwt laccessed 15 October 20031. Jung and Houde: Recruitment and spawning-stock biomass distribution of Anchoa mitchilli 65 Table 1 Cruise dates, mean standard errors for temperatures (°C) individual cruises. salinities (psul, and dissolved oxygen (mg/L). ir years, seasons, and regions of Chesapeake Bay, tegrated from surface to bottom, and pooled 1995-2000. CV = coefficient of variation for annual means. Temperature SE Salinity SE Oxygen SE Cruise date ( depart 28 Apr 95 ure) 13.88 0.11 15.01 0.42 8.53 0.1.3 23 Jul 95 28.13 0.12 15.48 0.44 6.50 0.14 28 Oct 95 17.26 0.12 17.39 0.45 7.59 0.14 28 Apr 96 13.87 0.10 10.84 0.36 10.21 0.11 17 Jul 96 24.66 0.11 11.80 0.41 7.43 0.13 22 Oct 96 16.10 0.10 11.26 0.36 8.55 0.11 20 Apr 97 10.93 0.13 11.41 0.50 10.01 0.16 11 Jul 97 25.28 0.13 13.59 0.51 7.10 0.16 29 Oct 97 14.64 0.13 18.19 0.51 8.01 0.16 11 Apr 98 12.26 0.12 8.90 0.44 9.95 0.14 04 Aug 98 26.15 0.12 12.89 0.46 7.01 0.15 19 Oct 98 18.60 0.13 16.64 0.49 8.64 0.15 19 Apr 99 11.97 0.13 13.51 0.49 10.04 0.16 26 Jun 99 23.52 0.15 16.02 0.56 5.75 0.18 23 Oct 99 16.30 0.14 17.38 0.53 8.87 0.17 29 Apr 00 12.95 0.17 12.51 0.64 8.98 0.20 25 Jul 00 24.26 0.14 14.06 0.53 5.17 0.17 17 Oct 00 17.89 0.15 16.73 0.56 7.63 0.18 Year 1995 19.76 0.07 15.96 0.25 7.54 0.08 1996 18.21 0.06 11.30 0.22 8.73 0.07 1997 16.95 0.08 14.40 0.29 8.37 0.09 1998 19.00 0.07 12.81 0.27 8.53 0.08 1999 17.26 0.08 15.64 0.30 8.22 0.10 2000 18.36 0.09 14.43 0.33 7.26 0.11 CV 5.8% 12.5% 1.2% Season April-May 12.64 0.05 12.03 0.20 9.62 0.06 June-August 25.33 0.05 13.97 0.20 6.49 0.06 October 16.80 0.05 16.27 0.20 8.22 0.06 Region of bay Lower 18.40 0.04 21.19 0.16 8.15 0.05 Middle 18.33 0.05 14.06 0.19 8.33 0.06 Upper 18.04 0.06 7.02 0.23 7.85 0.07 tions per survey ranged from 24 to 52 (six-year total=597). Additional baywide surveys (August 1997 and September 1998) and partial surveys (June 1997, July 1998, and July 1999) also provided data (total stations =146). An 18-m 2 mouth-opening midwater trawl (MWT), with 3-mm codend mesh was deployed from the stern of the 37-m research vessel Cape Henlopen. All trawling was conducted at night. Standardized tows of 20-min dura- tion were conducted and the trawl was deployed at graded depth intervals from surface to bottom ( 2 minutes at each depth interval ) in order to provide a sample of fish from the entire water column. Fish catches (or subsamples) were counted, measured (to the nearest 1.0 mm), and weighed on deck immediately after a tow. Abundance and biomass of bay anchovy recruits and spawners We separated bay anchovy catches into YOY and spawn- ers based on total length (TL). The minimum length of bay anchovy retained by the MWT was 21 mm TL, which we also defined as the minimum TL for recruited YOY bay 66 Fishery Bulletin 102(1) anchovy. Modal lengths of young-of-the-year (YOY) bay anchovy cohorts were determined from length-frequency distributions in MWT catches and a modal analysis (Bhat- tacharya, 1967; King, 1995). Based on the modal analysis of summer and fall survey data, the maximum TL of YOY bay anchovy and, therefore, the minimum TL of spawners, were estimated (Table 2). Length-dependent gear selectivity for bay anchovy was adjusted by comparing catches of the MWT and a 2-m 2 Tucker trawl with catches from 707-iim meshes at the same stations during a September 1998 baywide survey. The length-specific MWT:Tucker-trawl catch ratios (N^^j/ iVj^catch per unit of effort MWT 4- catch per volume of water Tucker trawl) for anchovies 21-70 mm TL indicated that both gears fished with a consistent selectivity for bay anchovy of 30-48 mm TL, and with a slight decrease in N TT for 48-70 mm TL. However, the values ofN MWT IN TT were lower by factors of 1-7 for 21-30 mm TL fish, indicating that small anchovies were collected less efficiently by the MWT. We concluded that length classes of anchovies >30 mm TL were equally vulnerable to the MWT and those >48 mm TL were less vulnerable to the Tucker trawl. Accord- ingly, we adjusted MWT catches of ^30 mm TL anchovy by multiplying them by a weighting factor estimated from the regression of values of iV MH , T /./V. r7 . for 21-30 mm TL bay anchovy. ( Weighting factor) = -0.59 TL + 19.08, (r 2 =0.96) where TL = total length. The weighting factor equals 1.0 for anchovy >30 mm TL because MWT selectivity is constant for anchovy >30 mm TL. To estimate water sampled in a 20-min MWT tow, and where D« d n = n mwt/ v mwt = ( 1/s ' x Nt/Vtt MWT — ^ x ^-^ MWT TT x TT bay N, MWT N 77' the concentration of 31-48 mm TL anchovy at a station (i.e. number/m 3 ); the number of 31-48 mm TL bay anchovy collected per 20-min MWT tow at a station; V MWT = the effective water volume sampled by a 20-min MWT tow (m :! ); the number of 3 1-48 mm TL bay anchovy col- lected by the 2-m- Tucker trawl at the same station; vulnerability to the Tucker trawl (s=l if all bay anchovies in water volume, V^, are col- lected); and V TT is the volume filtered by the Tucker trawl (m 3 ) estimated from a flowme- ter in its mouth. The mean of Af WHT /./V 7T for 30-48 mm TL bay anchovy during the September 1998 survey indicated that V' WUT = 4961 m\ if 30-48 mm TL bay anchovy did not significantly avoid the mouth of the 2-m 2 Tucker trawl (i.e. s=l). Assum- ing s=l (i.e. V MVVT =4961 m 3 ), we estimated "relative" bay- Table 2 Estimated maximum total lengths of young-of-the-year bay anchov y (mm ) from Chesapeake Bay, based on analy- sis of length-frequency distributions. Year Date Length (mm) 1995 23 Jul 28 Oct 52 69 1996 17 Jul 22 Oct 57 68 1997 11 Jul 2 Aug 29 Oct 30 56 66 1998 4 Aug 7 Sep 19 Oct 50 62 69 1999 26 Jun 23 Oct 30 65 2000 25 Jul 17 Oct 52 67 wide abundance and biomass of YOY and spawners for the 18 surveys from 1995 to 2000. To coarsely estimate a typical value of s. "absolute" bay- wide spawner biomasses in June— August were estimated for 1995-2000 according to an egg production method (Parker, 1985; Rilling and Houde, 1999a). Bay anchovy eggs had been collected in a 1-m 2 Tucker trawl during the same surveys and provided estimates of egg abundance. The coverage of stations and sampling design for the Tucker trawl was comparable to that of the MWT, but the Tucker trawl was deployed during both day and night. We presumed that all eggs collected between 00:00 and 20:00 h had been spawned near a midnight peak 1 00:00 h) (Za- strow et al., 1991) and decreased in abundance at a mean instantaneous mortality (reported for bay anchovy eggs in Chesapeake Bay as M = 0.066/h; Dorsey et al., 1996). Based on the estimated number of eggs spawned at 00:00 h for each station, the regional mean weight of individual spawners (defined by the minimum TL in Table 2) in MWT catches, and the reported fecundity-weight relationship for females (Zastrow et al., 1991), we were able to coarsely estimate "absolute" baywide spawner biomass. We as- sumed that the spawning fraction of adult females per day- was essentially 1.0 (i.e. all adult females participated in spawning, Zastrow et al., 1991) and the fecundity-weight relationship was constant over years. Comparison of the baywide estimates of spawner bio- mass in June-August based on the egg production method ("absolute" biomass) with estimates based on the MWT catch-per-unit-of-effort ("relative" biomass) indicated that. on average, for 1995 to 2000, s is equal to 0.20. Therefore, the mean effective water volume fished by a 20-min MWT tow was 4961x0.20 = 989 m 3 . Jung and Houde: Recruitment and spawning-stock biomass distribution of Anchoa mitchilli 67 Because N mvT of bay anchovy was highly variable, even at stations on the same sampling transect, and a mixed model (SAS version 6.12, SAS Inst. Inc., Cary, NC) includ- ing spatial covariance ( variogram ) did not significantly im- prove precision in annual, seasonal, and regional means or differences of N MWT , a stratified sampling design ( Steel and Torrie, 1980), i.e. stratum = region, was adopted. Based on the mean effective water volume (=sxV MWJ , ), we estimated regional "absolute" abundance and biomass (number and wet weight) and related standard errors of the linear com- bination by regional subvolumes (Samuels, 1989) of bay anchovy >21 mm TL for all MWT surveys from 1995 to 2000 by multiplying regional mean MWT catch by V r /989, where V r represents the water volume (m 3 ) in each bay region (Cronin, 1971): N !olal =(N^V l+ N n V,„ + N. Vj/(sxV MWT )xV lotal SE N =Sc N jVr/n,+V*/n n +v:?/n„ where N lotal v„ v m , v u SE X Sc N = baywide absolute abundance; mean values of N mvT for the lower (1), middle (m), and upper (u) bay; bay subvolumes for the lower (1), middle (m), and upper (u) bay (from Cronin, 1971), V, = 26.7 x 10 9 m 3 , V m = 16.8 x 10 9 m 3 , V„ = 8.7 x 10 9 m 3 , V„„„, = V, + V m + V„ =52.1 x 10 9 m 3 ; standard error of N lolal ; number of midwater trawl stations for the lower (1), middle (m), and upper (u) bay; pooled standard deviation of N MWT = square root of mean squares within groups in analysis of variance table = t/< SS, + SS m + SS„ ) / ( n, lMl -3i, where SS,, SS m , SS tl = sum of squares of N MWT for the lower (1), middle (m), and upper (u) bay, and "total = n l + n m + n u- Environmental factors Depth profiles of temperature, salinity, and dissolved oxygen ( DO ) concentration were determined from conduc- tivity-temperature-depth ( CTD ) casts at sampling stations. DO data were adjusted by calibrating against Winkler titration data from water samples collected in Niskin bot- tles deployed with the CTD cast. However, DO data from the CTD could not be adjusted for the 1999 summer and all calendar year 2000 cruises because Winkler titrations were not conducted. To estimate regional means for the water column, we averaged temperature, salinity, and DO values by integrating the observed values with respect to depth, after dividing the water column into "above pycnocline" and "subpycnocline" layers. Ontogenetic migration We analyzed length-frequency distributions along the south-north axis of the bay (i.e. by latitude) to delineate possible ontogenetic migrations of YOY and adult bay anchovy. To parameterize the distribution of YOY and adult abundance and biomass, we estimated the biomass- weighted mean latitudes of occurrence for each length class (3-mm interval). l b.i = 2_, B kjL k /2jB tl , where L B , = biomass-weighted mean latitude of a length class, /; L k = latitude of the station, k; and B = biomass (g, wet weight) per 20-min tow. We devised a metric to parameterize the location of bay anchovy SSB. We assumed that the baseline boundary for SSB distribution during the spring was at the mouth of the bay (37°00'N). Then, the upbay difference between biomass-weighted mean latitude of SSB (in decimal units) in Jun-August and the baseline for SSB during the spring lAL i was calculated: SL biomass-weighted mean latitude of SSB in June - August -37.00. Recruitment model As an exploratory step, a correlation analysis was under- taken to examine the relationships between bay anchovy SSB, migration patterns, and recruitment levels with respect to regional and depth-layer-specific mean tempera- ture, mean salinity, mean DO, their gradients, and monthly mean freshwater flow from the Susquehanna River. Cross- correlations revealed that SSB migration pattern {AD, regional mean DO concentrations, and October YOY recruitment level were closely correlated. Regional mean DO concentration provided the best fit to YOY recruitment level in October when baywide SSB also was included as an explanatory variable in multiple regressions. However, because there is uncertainty in the uncalibrated DO measurements in 1999 and 2000. we did not use regional mean DO in our recruitment model. Instead, we developed a modified Ricker-type stock-recruitment model (Ricker, 1975) that included AL as an explanatory variable: R x = a S exp (-/3j S - /i, AL) + e (modified Ricker model ) where R, recruitment level = October YOY abun- dance in each year ( 1995-2000); y; a, l\ and p.-, = regression coefficients; S = estimated baywide SSB (male-i- female) in metric tons for April-May; and £ = the error term. In this model, if AL is held constant, R s . is maximum at S = l//3j. Although no abiotic factor was included explicitly in the model, AL is strongly correlated with regional mean DO and serves as a proxy for it. For the modified Ricker model, collinearity, and jackknife influence diagnostic tools were 68 Fishery Bulletin 102(1) Table 3 Seasonal mean freshwater flow entering Chesapeake chesbay/RIMP/adaps.html. Bay ft' Dm the Susqu ehanna River ( m 3 /s ). Data source : http://va. water. usgs.gov/ Period 1995 1996 1997 1998 1999 2000 Jan-Mar 1289 2495 1474 2563 1325 1379 Apr-Jun 728 1702 920 1625 791 1627 Jul-Sep 238 768 239 334 294 393 Oct-Dec 923 2230 746 194 642 504 Annual mean 795 1799 845 1179 763 976 applied to evaluate reliability of the regression model (Belsley et al„ 1980; SAS, 1989). Results Environmental factors Stream flows from the Susquehanna River (Table 3) varied annually and seasonally. Freshwater stream flows were higher in 1996 and 1998 than in other years. Baywide mean values of water temperature, salinity, and DO concentration, averaged from surface to bottom, varied annually, seasonally, and regionally (Table 1 ). Annually, mean temperature was highest in 1995 and lowest in 1997. Mean salinity was highest in 1995 and lowest in 1996. Mean DO concentration was highest in 1996 and lowest in 2000. Regionally, salinity was more variable than temperature and DO concentration. Seasonally, temperature and DO concentration were more variable than salinity. Tem- perature was highest in the June-August period, the spawning season of bay anchovy. Seasonally, salinity increased progressively from April-May to October. Mean DO concentration was consistently lowest in June-August. Trends in abundance and recruitment Estimates of bay anchovy abundance reported in our study are for the entire mainstem of Chesapeake Bay. The estimated recruitment levels (baywide abundance of YOY bay anchovy >30 mm TL in October) varied ninefold and were low in 1995 and 1996 (47.5 ±16.6 and 30.6 ±8.6xl0 9 individuals) but much higher in 1997-2000 (99.6 ±12.4 to 264.8 ±32.6xl0 9 ). Baywide estimates of bay anchovy biomass for individuals >30 mm TL increased from April to October in each year (Table 4). October baywide biomass varied sevenfold from 27.1 ±5.5 x 10 3 to 192.9 ±20.4 x 10 3 tons and was highest in 1998 and lowest in 1996. Estimated spawning stock biomass (SSB) in April-May was lowest in 1995 (3.3 ±1.1 x 10 3 tons), and highest in 1997 (20.1 ±5.3 x 10 3 tons), indicating sixfold variability. SSB in June-August was lowest in 1996 (2.4 ±0.2 x 10 3 tons), and highest in 1997 (21.1 ±2.3 x 10 3 tons). The SSBs in April-May and June-August did not show any obvious relationship to YOY abundance (recruitment) in October. Ontogenetic migration The length-specific mean locations (latitudes of occur- rence ) of bay anchovy revealed an apparent ontogenetic migration. Small juveniles of bay anchovy tended to move upbay and were located primarily upbay until they were approximately 45 mm TL, after which they began to move downbay (Fig. 2). In April-May, age-1 bay anchovy <60 mm TL, consisting of individuals recruited from the previous year, varied annually in their mean latitude of occurrence, whereas large (sage 1, a60 mm TL) bay anchovy had relatively stable locations near the boundary between the lower and middle bay regions, centered at latitude 37°40'N (Fig. 2A). Compared to April-May, age-l+ bay anchovy in June-August were more variable in their annual mean locations, but both YOY and adult bay anchovy tended to occur upbay of latitude 38°00'N, except in year 2000 (Fig. 2B). In 1997 and 1999, when annual mean temperatures were lowest (Table 1), YOY bay anchovy were too small to be sampled by the MWT in June-August and are not represented in Figure 2B. In October, mean latitudes of occurrence (Fig. 2C) indicated a consistent distribution pattern and an apparent ontogenetic migration by YOY anchovy. The most probable explanation for the observed latitudinal distributions was that small YOY bay anchovy tended to move upbay initially, but then downbay at about 45 mm TL. Distribution of age-l-t- individuals in October was variable. The SSB of bay anchovy (excludes YOY) from 1995 to 2000 was centered near 38°00'N in April-August except in June-August of 1995 and 1996, when the SSB was centered farther upbay (Fig. 3A). In 2000, the migration pattern differed from other years. Spawning bay anchovy in 2000 were located farther downbay in July than in April (Fig. 3A). The April-May location of prespawning SSB was mostly explained by the mean flow of the Susquehanna River from June of the previous year to February of the current year (r 2 =0.94, P=0.0012; Fig. 3B ). But, in June-Au- gust, the mean location of spawning fish was more strongly and significantly related to the subpycnocline-layer mean Jung and Houde: Recruitment and spawning-stock biomass distribution of Anchoa mitchilli 69 Table 4 Baywide abundance and biomass estimates for bay anchovy >30 mm TL (young-of-the-year + adult). SE = = standard error. Year Period Abundance I xlO 9 ) Biomass xlO 3 metric tons) Estimate SE Estimate SE 1995 April-May 2.1 0.7 3.3 1.1 June-August 57.8 28.1 32.6 17.5 October 47.5 16.6 51.9 21.0 1996 April-May 4.9 1.1 8.9 2.0 June-August 5.3 1.6 3.7 1.3 October 30.6 8.6 27.1 5.5 1997 April-May 11.8 3.3 20.1 5.3 June-August 9.4 2.3 21.1 5.0 October 99.6 12.4 85.6 10.8 1998 April-May 3.5 0.7 6.1 1.3 June-August 14.4 4.5 17.0 7.9 October 264.8 32.6 192.9 20.4 1999 April-May 6.9 1.4 10.6 2.2 June-August 5.5 1.2 10.6 2.4 October 124.5 28.3 115.3 25.0 2000 April-May 6.2 4.1 13.0 6.6 June-August 144.6 51.2 56.0 17.0 October 169.1 43.7 152.9 40.0 DO during that same period in the middle bay (/•-=(). 75, P=0.02;Fig. 3C). Correlations Correlation analyses suggested that regional mean DO concentrations are the most important environmental correlate associated with spatial distribution of SSB and recruitment processes of bay anchovy. The mean locations (latitudes of occurrence), abundances, and biomasses for YOY and adult bay anchovy were analyzed with respect to environmental variables (Table 5). Recruitment levels (YOY abundance) in October were consistently inversely correlated with DO concentrations in the lower and middle bay in June-August (/-=-0.13 to -0.89). Biomass- weighted mean latitude of SSB (age 1+) in April-May was consistently and positively correlated with regional salini- ties in April-May (r=0.30 to 0.88). On the other hand, in June-August, surface-layer mean salinity in the lower Bay and subpycnocline-layer mean DO in the lower and middle bay were significantly and positively correlated with mean latitude of SSB or AL (r=0.82 to 0.91). Baywide SSB in April-May and June-August tended to be negatively cor- related with water temperature in April-May (r=-0.45 to -0.90). Recruitment model Although SSB alone did not correlate significantly with recruitment level, mean DO in June-August was signifi- cantly related to the mean latitude of SSB in June-August (or AL) and bay anchovy recruitment level in October (Figs. 3C and 4). AL was selected as the explanatory variable, rather than DO, because DO data were uncalibrated in 1999 and 2000. The correlation observed between AL and DO ( Fig 3C ) suggested that AL can serve as a proxy for DO in the stock-recruitment model. Including AL and SSB for April-May in a modified Ricker model provided a good fit to bay anchovy recruitment levels observed from 1995 to 2000 (Fig. 5). The model is R v = 365 S exp (-0.19 S 1.35 AL) (modified Ricker model). In the model, if AL is held constant, predicted recruitment level of bay anchovy is maximum when baywide SSB in April-May is approximately 5.3 x 10 3 tons. Collinearity and influence diagnostic statistics did not indicate collinearity between the two independent variables (S and AL), or that an observation in any year had a dominating influence on parameter estimates. Discussion Complex environmental processes and biological interac- tions control bay anchovy recruitment in Chesapeake Bay. Dissolved oxygen (DO), freshwater flow, salinity, and tem- perature acting on prerecruits and adults are important factors affecting bay anchovy distribution and levels of recruitment. Spawning stock size also is related to recruit- 70 Fishery Bulletin 102(1) ment level. Our results have demonstrated that there is a strong spatial component in the recruitment dynamics of bay anchovy. Although fish recruitment processes his- torically have been difficult to understand, our six-year, spatially extensive research has provided new insights into processes that control bay anchovy recruitment. Ontogenetic migration pattern It is apparent that ontogenetic migration plays a role in the spatial and temporal patterns in abundance, biomass, and production of bay anchovy. There are several lines of evidence. Rilling and Houde (1999a), in a baywide analy- sis, reported that mean density of eggs and larvae in June and July 1993 was very high in the lower Chesapeake Bay compared to more upbay sites. Dovel (1971) and Loos and Perry (1991) reported possible upbay or upriver migra- tion of bay anchovy larvae and juveniles in the mainstem and tributaries of the Bay. Recent otolith microchemical analyses have strongly supported the hypothesis that an upbay ontogenetic migration by small YOY anchovy (>25 mm, late larvae and small juveniles) occurs (Kimura et al., 2000). In the middle Hudson River estuary (Schultz April-May 39°00' £ 38°00 37°00 39°00 38°00 37°00' 30 40 50 60 70 80 90 100 TL (mm) 1995 1996 1997 1998 1999 2000 Figure 2 Abundance-weighted mean latitude of occurrence of bay anchovy (Am hoa mitchilli) in Chesapeake Bay, 1995-2000. et al., 2000) and Chesapeake Bay (North and Houde, in press), selective tidal-stream transport was suggested as a mechanism for up-estuary movements of bay anchovy larvae. Our conceptual model of the bay anchovy life cycle includes migration patterns in the bay based on available knowledge and evidence (Fig. 6). It is uncertain what benefits YOY bay anchovy derives from upbay migration in summer and whether the migra- tion is passive or active before a subsequent reverse migra- tion in the fall. To explain upbay movements of estuarine fishes, Dovel ( 1971 ) proposed that there is a "critical zone" of low salinity and high prey production in the upper bay, which is important as a nursery for bay anchovy and other fish species. In late spring and early summer, age-1 and age-l+ bay anchovy mature and move upbay while spawning, although the year 2000, when mean freshwater streamflow during the previous fall-winter was lowest, was an exception. Recruited YOY bay anchovy apparently over- winter primarily, but not entirely, downbay until spring. There remains a possibility of significant immigration to the bay by adult bay anchovy in some years from the coastal ocean or tidal tributaries of the bay. Without such immigration, baywide adult abundance would decrease continuously during the April-October period through natural mortality However, in two years of our six-year study, 1995 and 1998, estimated adult abundance in- creased substantially from April to July, and in 1999 adult abundance increased from June to October, implying significant immigration to the bay in those years (Jung, 2002). Recruitment control and regulation The modified Ricker recruitment model that included SSB and AL as explanatory variables provided a good fit to bay anchovy recruitments. Although the model fitted well, there were only six years of data, and the underlying mechanisms explaining relationships between the distribution and level of SSB, hydro- logical conditions, and density-dependent regulatory processes in recruitment of bay anchovy are not yet clear. Nevertheless, correlations and the recruitment model clearly indicated a density-dependent effect of SSB level and also implicated environmental factors (at the mesoscale) that are related to mean DO concen- tration, latitudinal distribution of SSB (AL), and the recruitment level of bay anchovy (Fig. 4). The modified Ricker model for bay anchovy < Fig. 5) indicates a density-compensatory stock-recruitment relationship (Ricker, 1975). although we do not know at what life stages density-dependent processes are most important. Without accounting for the control- ling effect of AL and mean DO on a regional scale, the density-dependence might have gone undetected (Fig. 4 1. Recent individual-based models suggest that density-dependent processes during early-life stages could stabilize bay anchovy recruitments (Wang et al., 1997; Cowan et al., 1999; Rose et al, 1999). At the small scales of several meters modeled by Wang et al. (1997) and Cowan et al. (1999), larval-stage feeding Jung and Houde: Recruitment and spawning-stock biomass distribution of Anchoa mitchilli 71 Zl 39°00' 38°00' CO CO CO 37°00' April-May June-August 1995 1996 1997 1998 1999 2000 2000 38°00' 1999^ 1=38.30 - 0.00087 X r-=0.94(/)=0.0012) 37045' 1995 --4?^- 1998 B 1996_ 37°30' i , 300 < 39°00' c C CO CO c 38°00' 400 500 600 700 Mean river flow from June to Feb (m 3 /sec) 1995 1996^ 1999 1998 37°00' 2000 1997 Y= 35.78 + 0.53 A' r 2 =0.75(p=0.02) 3.0 3.5 4.0 4.5 5.0 Dissolved oxygen (mg/L) 5.5 Figure 3 Mean location (latitude) of adult bay anchovy {Anchoa mitchilli) spawn- ing stock biomass (SSB) in Chesapeake Bay. (A) Mean latitude and standard deviation in April-May and in June-August. The upper verti- cal bar represents mean + standard deviation for June-August, and the lower vertical bar represents mean-standard deviation for April-May, I B l Mean latitude in April-May and mean Susquehanna River flow from June of the previous year to February of the current year. (C) Mean lati- tude in June-August and mean dissolved oxygen in the subpycnocline layer of the middle bay in June-August. processes were important and high adult SSB could pro- duce abundant first-feeding larvae with subsequent den- sity-dependent food competition. In Tampa Bay, Florida, Peebles et al. ( 1996) hypothesized that bay anchovy's size- specific fecundity is directly related to prey availability for adults. Modeled results of Rose et al. (1999) suggested that density-dependent growth of bay anchovy larvae and juveniles in Chesapeake Bay would lead to density-depen- dent survival of these stages. Hunter and Kimbrell (1980) and Alheit (1987) proposed that cannibalism by adults on eggs and larvae provides a degree of density-dependent regulation in anchovies of the genus Engraulis. Analyses of feeding by adult bay anchovy did not indicate that pe- lagic fish eggs were a significant part of bay anchovy diet (Vazquez-Rojas, 1989; Klebasko, 1991), although no specific study of cannibalism has been undertaken. We propose three hypotheses that may explain the rela- tionships among regional DO concentration, the latitudi- nal shift in SSB distribution during the spawning season (AL), and recruitment levels of bay anchovy in October. The 72 Fishery Bulletin 102(1) hypotheses are the following: 1) averaged DO concentra- tion is inversely related to levels of plankton productivity in a region and high plankton productivity favors high re- cruitments of planktivorous bay anchovy; 2 ) low dissolved oxygen concentrations can restrict spatial distribution of bay anchovy SSB to the lower bay insuring high egg and Table 5 Cross-correlation coefficients for bay anchovy distribution and abundance with respect to region- and layer-specific means of tem- perature, salinity, and dissolved oxygen from 1995 to 2000. Mean latitude is biomass-weighted mean latitude of occurrence of bay anchovy. Abundance and biomass are baywide total estimates. AL = (mean latitude in June-August) -37.00. Abbreviations are as follows: SAL = salinity, TEM = water temperature, OXY = dissolved oxygen; the fourth and fifth digits: 04 = April-May, 07 = June-August; the sixth character: L = lower bay, M = middle bay, U = upper bay; The last character: S = layer above the pycnocline. B = layer below the pycnocline. * = significant at a = 0.05. Young-of-the-year Adult Mean latitude Abundance Mean latitude Biomass April-May June-August (orAL) October October April-May June-August SAL04LS 0.29 -0.43 0.74 0.26 -0.17 -0.52 SAL04MS 0.45 -0.63 0.30 0.71 -0.41 -0.22 SAL04US 0.27 -0.60 0.42 0.53 -0.18 -0.02 SAL04LB -0.24 0.01 0.88* -0.16 -0.14 -0.31 SAL04MB 0.08 -0.17 0.59 0.33 -0.39 -0.05 SAL04UB 0.29 -0.61 0.45 0.46 -0.03 0.05 SAL07LS 0.83* -0.75 0.91* -0.46 SAL07MS -0.12 0.06 0.14 0.31 SAL07US 0.06 -0.03 -0.04 -0.33 SAL07LB 0.70 -0.75 0.64 -0.11 SAL07MB -0.41 0.60 -0.31 0.19 SAL07UB 0.15 -0.20 0.01 -0.42 TEM04LS 0.16 -0.25 -0.03 0.65 -0.90* -0.48 TEM04MS 0.50 -0.46 0.14 0.65 -0.71 -0.85* TEM04US 0.53 -0.32 -0.36 0.52 -0.56 -0.85* TEM04LB 0.29 -0.49 0.19 0.71 -0.72 -0.45 TEM04MB 0.22 -0.42 0.39 0.47 -0.55 -0.62 TEM04UB 0.40 -0.26 -0.39 0.48 -0.60 -0.77 TEM07LS -0.49 -0.04 0.11 0.45 TEM07MS -0.16 -0.21 0.47 0.14 TEM07US -0.29 -0.08 0.39 0.38 TEM07LB -0.68 0.24 -0.11 0.38 TEM07MB -0.24 -0.10 0.37 -0.04 TEM07UB -0.45 0.16 0.2] 0.46 OXY04LS 0.63 -0.22 -0.80 0.39 -0.10 -0.30 OXY04MS -0.27 0.56 0.23 -0.81 0.55 -0.04 OXY04US -0.43 0.41 -0.30 -0.30 0.30 0.88* OXY04LB 0.93** -0.68 -0.59 0.63 0.04 -0.38 OXY04MB 0.47 -0.35 -0.31 -0.09 0.70 -0.12 OXY04UB -0.57 0.65 -0.32 -0.46 0.21 0.78 OXY07LS 0.18 -0.30 0.29 0.32 OXY07MS 0.01 -0.13 0.29 0.56 OXY07US 0.23 -0.32 0.50 0.10 OXY07LB 0.67 -0.48 0.82* -0.28 OXY07MB 0.72 (l,SH 0.87* -0.04 OXY07TJB 0.01 0.16 0.21 0.37 Jung and Houde: Recruitment and spawning-stock biomass distribution of Anchoa mitchilli 73 larval production there; and 3) density-depensatory predator satiation occurs when concentrations of bay anchovy larvae and juveniles at the mesoscale ( 10-100 km ) are high in relation to satiation potential of preda- tors, which favors larval production and high anchovy recruitments. First, averaged DO level in the bay or its regions may be an indicator of ecosystem metabolism and sec- ondary production. DO level in the subeuphotic layer is an indicator of respiration and secondary produc- tion by planktonic and benthic communities (Kemp and Boynton, 1980; Kemp et al., 1992). Recruitment levels of bay anchovy increased substantially in 1997 and in subsequent years. We speculate that enhanced detrital production potentially increased zooplankton prey abundances in the subsequent year and that asso- ciated elevated levels of respiration by detrital micro- organisms and zooplankton contributed to low mean DO. Increased zooplankton prey abundances, in turn, may have promoted production of larval and juvenile bay anchovy in 1997 and 1998. Thus, increased prey availability, associated with low mean DO concentra- tion, could have enhanced recruitment (Fig. 4). The second hypothesis proposes that spatial restric- tion of SSB by low DO is a factor controlling bay anchovy recruitment. Based on our results, hypoxic conditions in the bay appear to define the distribution and potential for upbay migration of bay anchovy SSB (Fig. 3C). In years 300 1998 7= -88 .V+ 5 10 C 200 o 1 00 cr r-=0.79P=0.01S 2000 ^~"-\1999 ^^19,97 ~"\J995 1W(, 3.0 3.5 4.0 4.5 5.0 Dissolved oxygen (mg/L) Figure 4 Relationship between mean dissolved oxygen below the pycno- cline in the middle Chesapeake Bay during the June-August period and recruitment level of bay anchovy in October, r 2 = coefficient of determination. when the baywide subpycnocline mean DO level was low, spawning bay anchovy tended to be most concentrated in the lower bay (Table 5, Fig. 3, A and C), possibly because hypoxia in deeper waters of the mid-bay region discouraged upbay migration. The region selected by adult anchovy as the predominant spawning area and its variability played R = 365 Sexpf-O.l - S - 1.354Z.) r 2-- 2.0 Figure 5 Stock-recruitment model (modified Ricker model). R = baywide number of recruits in October (xlO 9 ). AL = location of bay anchovy iA?iclioa mitchilli) spawning stock biomass in June-August in relation to the baseline latitude at the mouth of the bay, 37°00'N. S = baywide spawning stock biomass (SSB xlO 3 metric tons for April-May 1. Balloon symbols are observed data from 1995 to 2000. 74 Fishery Bulletin 102(1) a strong role in controlling YOY recruitment levels. The four highest recruitment years in our series had the lowest mean subpycnocline DO levels and had distribution pat- terns of SSB that differed little between the prespawning April-May and spawning June-August periods (Fig. 4). Al- though we do not fully understand how DO, and possibly hypoxic conditions, affect migratory behavior and distribu- tion patterns of bay anchovy, hypoxia in Chesapeake Bay has been demonstrated in other research to affect spatial and temporal patterns of fish abundance, including bay anchovy (Breitburg, 1992; Keister et al., 2000). Our third hypothesis proposes that predation is an im- portant regulator of fish recruitment in early-life stages (Sissenwine, 1984; Bailey and Houde, 1989). We hypoth- esize that abundant and spatially concentrated larval or juvenile anchovy, as observed in the lower bay, could promote early-life survival by satiating predators, even if some predators migrate to areas where larval and juvenile anchovy are abundant. At mesoscale distances of 10-100 km, distribution of predators (e.g. YOY and age-1 weakfish [Cynoscion regalis] ) may be important. If the maximum number of prey that can be eaten by predators is reason- ably constant, the effect of predation can be density-depen- satory (Hilborn and Walters, 1992), i.e. predation mortality rate decreases as prey density increases. In support of the third hypothesis, a correspondence analysis on fish species assemblages by year, season, re- gion, and life stage (Jung and Houde, 2003) indicated that distributions and abundances of YOY weakfish, a major predator of bay anchovy in Chesapeake Bay (Hartman and Brandt, 1995), and YOY bay anchovy were closely as- sociated spatially, seasonally, and annually in our six-year study. The major spawning area of bay anchovy is spatially restricted. If predator migration to the area is limited, then as the supply of larvae and juveniles increases, it may satu- rate predator demand, the condition necessary for depensa- tion to be important. It may seem contradictory to propose that density-com- pensation with respect to SSB (the negative sign of j\) and density-depensation with respect to AL (the second or third hypothesis ) can act simultaneously during larval and juvenile stages. Under this circumstance, the number of surviving postlarval anchovies is hypothesized to decrease because of food limitation when larval abundance is high, reducing subsequent predation-related mortality rate on postlarvae and small juveniles. Low abundance of anchovy early-life stages will lead to the opposite effect (Fig. 7). The proposed opposing responses of the early-larval and late- larval-juvenile stages are explained by differences in the spatial scales of distribution and densities of life stages of bay anchovy (Fig. 7). The spatial scale of processes that affect distributions of late-stage larvae and juveniles is large compared to that for early-stage larvae because of the increased dispersal and swimming ability of juveniles. Comparing early-larval and late-larval-juvenile stages of bay anchovy, we propose that effects of prey concentration (the first hypothesis) and SSB level (density-compensa- tion) act primarily on the dynamics of early-larval stages, whereas predation mortality and the inhibitory effects of low DO (density-depensation; the second and third hy- Nursery Ground (3) Fall YOY recruits, adults Late-stage larvae, juveniles, some adults Eggs and larvae Overwintering Recruited anchovy Adult Immigration from tributaries'? Major Spawning Mature adults. / eggs, larvae ground (1) Spring Adult Immigration from ocean? Figure 6 Conceptual model representing bay anchovy (Anchoa mitchilli) life cycle and ontogenetic migration within Chesapeake Bay, and possible immigration of adults from tributaries and coastal ocean. potheses) are more important regulators and controllers, respectively, during late-larval and juvenile stages. The three hypotheses that relate DO, SSB distribution, and recruitment of bay anchovy are not mutually exclusive. If low mean DO level is an indicator of enhanced prey pro- duction and availability to larvae and juveniles, increased prey productivity in the lower bay could enhance bay anchovy recruitment potential by supplying enough zoo- plankton prey to spawning adults, larvae, and juveniles. At the same time, low mean DO in the mid-Bay could confine most spawning bay anchovy to the lower bay. thus increas- ing spawning and larval production there, and possibly enhancing survival of juveniles by predator satiation. Ul- timately, other hypotheses may provide better explanations of the relationships between regional mean DO. latitudinal shifts in distribution of spawners, abundances of spawners. and recruitment of bay anchovy. For example, abundant gelatinous organisms, such as the scyphomedusa (Chn'sa- ora quinquecirra) and the lobate ctenophore \Mnemiopsis leidyi), can be important predators on early-stage anchovy and competitors with juveniles and adults (Purcell et al., Jung and Houde: Recruitment and spawning-stock biomass distribution of Anchoa mitchil/i 75 Early-stage and larvae Density-compensatory Prey is smaller Small scale (1 m-10 km) Densiy of early-stage larvae (1 m-10 m scale) Late-stage larvae and juveniles Density-depensatory Predator is bigger Mesoscale(IO-lOOkm) Recruits Ontogenetic migration Densiy of late-stage larvae (10-100 km scale) SSB Figure 7 Hypotheses and conceptual model of the bay anchovy {Anchoa mitchilli) recruitment process in Chesapeake Bay. The density-compensatory process acts at a small spatial scale during the early- larval stages, whereas the density-depensatory process acts at a broader spatial scale during late-stage larval and juvenile stages. The ontogenetic migration is controlled by dissolved oxygen levels and other hydrological factors. 1994), but their potential role with respect to bay anchovy recruitment could not be defined in our study. For the present, it is clear that most spawning occurs in the lower and mid Chesapeake Bay, from which larval and juvenile anchovies disperse upbay. We hypothesize that food avail- ability is the major factor controlling production of bay anchovy early-larval stages whereas predation becomes more important during late-larval and juvenile stages. Our results and hypotheses implicate density-related pro- cesses, operating at different spatial scales, as regulators of recruitment of bay anchovy in Chesapeake Bay. Acknowledgments We thank S. Leach, E. North, J. Hagy, C. Rilling, J. Cleve- land, A. Madden, D. O'Brien, B. Pearson, D. Craige, T. Auth, and the able crew of RV Cape Henlopen for assistance in field surveys. T. Miller and E. Russek-Cohen provided com- ments and assistance on statistical analyses. This research was supported by a U.S. National Science Foundation, Land Margin Ecosystem Research (LMER) program grant, "Trophic interactions in estuarine systems (TIES)" (grant DEB94-12113). Additional support was provided by NSF Grant OCE-9521512 and by National Oceanic and Atmo- spheric Administration grant NOAA, NA170P2656. Literature cited Able, K. W„ and M. P. Fahay. 1998. The first year in the life of estuarine fishes in the Middle Atlantic Bight. 342 p. Rutgers Univ. Press, New Brunswick, NJ. Alheit, J. 1987. Egg cannibalism versus egg predation: their signifi- cance in anchovies. S. Afr. J. Mar. Sci. 5:467-470. Bailey. K. M., and E. D. Houde. 1989. Predation on eggs and larvae of marine fishes and the recruitment problem. Adv. Mar. Bio. 25:1-83. Baird, D.. and R. E. Ulanowicz. 1989. The seasonal dynamics of the Chesapeake Bay ecosystem. Ecol. Monogr. 59:329-364. Belsley, D. A., E. Kuh, and R. E. Welsch. 1980. Regression diagnostics: identifying influential data and sources of collinearity, 292 p. John Wiley and Sons, Inc., New York, NY. Bhattacharya, C. G. 1967. A simple method of resolution of a distribution into Gaussian components. Biometrics 23:115-135. Breitburg, D. L. 1992. Episodic hypoxia in Chesapeake Bay: Interacting effects of recruitment, behavior, and physical disturbance. Ecol. Monogr. 62:525-546. Cowan, J. H., K. A. Rose, E. D. Houde, S. B. Wang, and J. Young. 1999. Modeling effects of increased larval mortality on bay anchovy population dynamics in the mesohaline Chesa- peake Bay: evidence for compensatory reserve. Mar. Ecol. Prog. Ser. 185:133-146. Cronin. W. B. 1971. Volumetric, areal, and tidal statistics of the Chesa- peake bay estuary and its tributaries, 15 p. Chesapeake Bay Institute, Johns Hopkins Univ., Baltimore, MD. Dorsey, S. E., E. D. Houde. and J. C. Gamble. 1996. Cohort abundances and daily variability in mortality of eggs and yolk-sac larvae of bay anchovy, Anchoa mitchilli, in Chesapeake Bay. Fish. Bull. 94:257-267. 76 Fishery Bulletin 102(1) Dovel, W. L. 1971. Fish eggs and larvae of the upper Chesapeake Bay. Chesapeake Biological Laboratory, special report 4, 71 p. Natural Resource Institute, Univ. Maryland. College Park, MD. Harding, L. W., M. E. Mallonee. and E. S. Perry. 2002. Toward a predictive understanding of primary produc- tivity in a temperate, partially stratified estuary. Estuar. Coast. Shelf Sci. 55:437-463. Hartman. K. J., and S. B. Brandt. 1995. Predatory demand and impact of striped bass, blue- fish, and weakfish in the Chesapeake Bay: applications of bioenergetics models. Can. J. Fish. Aquat. Sci. 52: 1667-1687. Hilborn, R.. and C. J. Walters. 1992. Stock and recruitment. In Quantitative fisheries stock assessments: choice, dynamics, and uncertainty (R. Hilborn and C. J. Walters, eds. ). p. 241-296. Chapman & Hall, New York. NY. Houde, E. D. 1987. Fish early life dynamics and recruitment variability. Am. Fish. Soc. Symp. 2:17-29. Houde, E. D., and C. E. Zastrow. 1991. Bay anchovy. In Habitat requirements for Chesa- peake Bay living resources, 2nd ed. (S. L. Funderburk, J. A. Mihursky, S. J. Jordan, and D. Riley, eds.), p. 8.1 to 8.14. Living Resources Subcommittee, Chesapeake Bay Program, Annapolis, MD. Hunter, J. R., and C. A. Kimbrell. 1980. Egg cannibalism in the northern anchovy, Engraulis mordax. Fish. Bull. 78:811-816. Jung, S. 2002. Fish community structure and the spatial and tem- poral variability in recruitment and biomass production in Chesapeake Bay. Ph.D. diss., 349 p. Univ. Maryland, College Park, MD. Jung, S., and E. D. Houde. 2003. Spatial and temporal variability of pelagic fish com- munity structure and distribution in Chesapeake Bay, U.S.A. Estuar. Coast. Shelf Sci. 58:335-351. Keister, J. E., E. D. Houde, and D. L. Breitburg. 2000. Effects of bottom-layer hypoxia on abundances and depth distribution of organisms in Patuxent River, Chesa- peake Bay. Mar. Ecol. Prog. Ser. 205:43-59. Kemp, W. M., and W. R. Boynton. 1980. Influence of biological and physical processes on dis- solved oxygen dynamics in an estuarine system: implica- tions for measurement of community metabolism. Estuar. Coast. Mar. Sci. 11:407-431. Kemp, W. M., P. A. Sampou. J. Garber. J. Tuttle. ami W. R. Boynton. 1992. Seasonal depletion of oxygen from bottom waters of Chesapeake Bay: roles of benthic and planktonic respira- tion and physical exchange processes. Mar. Ecol. Prog. Ser. 85:137-152. Kimura, R., D. H. Secor, E. D. Houde, and P. M. Piccoli. 200D Cp-estuar\ dispersal ofyoung-of-the-year bay anchovy Anchoa mitchilli in the Chesapeake Bay: inferences from microprobe analysis of strontium in otoliths. Mar. Ecol. Prog Ser 208:217-227. King, M. 1995. Fisheries biology, assessment and management, 341 p. Fishing News Books, Oxford. Klebasko. M.J. 1991. Feeding ecology and daily ration of bay anchovy. Anchoa mitchilli in the mid-Chesapeake Bay. M.S. thesis, 103 p. Univ. Maryland. College Park, MD. Loos, J. J., and E. S. Perry. 1991. Larval migration and mortality rates of bay anchovy in the Patuxent River. In Larval fish recruitment and research in the Americas: proceedings of the 13th annual fish conference, 21-26 May 1989, Merida. Mexico (R. D. Hoyt, ed.), p. 65-76. NOAA Technical Report NMFS 95. Luo, J., and S. B. Brandt. 1993. Bay anchovy Anchoa mitchilli production and con- sumption in mid-Chesapeake Bay based on a bioenergetics model and acoustic measurement offish abundance. Mar. Ecol. Prog. Ser. 98:223-236. Luo, J., and J. Musick. 1991. Reproductive biology of the bay anchovy in Chesa- peake Bay. Trans. Am. Fish. Soc. 120:701-710. MacGregor, J. M., and E. D. Houde. 1996. Onshore-offshore pattern and variability in distribu- tion and abundance of bay anchovy Anchoa mitchilli eggs and larvae in Chesapeake Bay. Mar. Ecol. Prog. Ser. 138: 15-25. Murdy, E. O, R. S. Birdsong, and J. A. Musick. 1997. Fishes of Chesapeake Bay, 324 p. Smithsonian Insti- tution Press, Washington, D.C. and London. Newberger, T. A., and E. D. Houde. 1995. Population biology of bay anchovy Anchoa mitchilli in the mid Chesapeake bay. Mar. Ecol. Prog. Ser. 116:25-37. North, E. W„ and E. D. Houde. In press. Transport and dispersal of bay anchovy {Anchoa mitchilli) eggs and larvae in Chesapeake Bay. Estuar. Coast. Shelf Sci.. Parker, K. 1985. Biomass model for the egg production method. In An egg production method for estimating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax (R. Lasker, ed. ). p. 5-7. NOAA Tech. Rep. NMFS 36. Peebles, E. B.. J. R. Hall, and S. G Tolley. 1996. Egg production by the bay anchovy Anchoa mitchilli in relation to adult and larval prey fields. Mar. Ecol. Prog. Ser. 131:61-73. Purcell, J. E., D. A. Nemazie, S. E. Dorsey, E. D. Houde, and J. C. Gamble. 1994. Predation mortality of bay anchovy Anchoa mitchilli eggs and larvae due to scyphomedusae and ctenophores in Chesapeake Bay. Mar. Ecol. Prog. Ser. 1 14 :47-58. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics of fish population. Bull. Fish. Res. Board Can. 191: 1-382. Rilling, G. C, and E. D. Houde. 1999a. Regional and temporal variability in distribution and abundance of bay anchovy (Anchoa mitchilli i eggs, larvae, and adult biomass in the Chesapeake Bay. Estuaries 22: 1096-1109. 1999b. Regional and temporal variability in growth and mortality of bay anchovy. Anchoa mitchilli. larvae in Chesa- peake Bay. Fish. Bull. 97:555-569. Rose, K. A.. J. H. Cowan, M. E. Clark. E. D. Houde. and S. B. Wang. 1999. An individual-based model of bay anchovy population dynamics in the mesohaline region of Chesapeake Bay. Mar. Ecol. Prog. Ser. 185:113-132. SAS Institute Inc. 1989. SAS/STAT user's guide, version 6, 4th ed., 1686 p. SAS Institute Inc.. Gary, NC. Jung and Houde: Recruitment and spawning-stock biomass distribution of Anchoa mitchilli 77 Samuels, M. L. 1989. Statistics for the life sciences, p. 409-42. Prentice- Hall. Inc., Upper Saddle River, NJ. Schultz. E. T., R. K. Cowen, K. M. M. Lwiza. and A. M. Gospodarek. 2000. Explaining advection: Do larval bay anchovy lAnchoa mitchilli) show selective tidal-stream transport? ICES J. Mar. Sci. 57:360-371. Sissenwine, M. P. 1984. Why do fish populations vary? In Exploitation of marine communities (R. M. May, ed.l, p. 59-94. Springer- Verlad, Berlin. Smith, E. M., and W. M. Kemp. 2001. Size structure and the production/respiration balance in a coastal plankton community. Limnol. Oceanogr. 46: 473-485. Steel, R. G. D., and J. H. Torrie. 1980. Principles and procedures of statistics. A biometrical approach. 2 nd ed., 633 p. McGraw-Hill Inc. New York, NY. Vazquez-Rojas, A. V. 1989. Energetics, trophic relationships and chemical compo- sition of bay anchovy, Anchoa mitchilli in the Chesapeake Bay. M.S. thesis, 166 p. Univ. Maryland, College Park, MD. Wang, S. B., J. H. Cowan, K. A. Rose, and E. D. Houde. 1997. Individual-based modelling of recruitment variability and biomass production of bay anchovy in mid-Chesapeake Bay. J. Fish Biol. 51 (suppl. A):121-134. Wang, S. B., and E. D. Houde. 1995. Distribution, relative abundance, biomass and produc- tion of bay anchovy Anchoa mitchilli in the Chesapeake Bay. Mar. Ecol. Prog. Ser. 121:27-38. Zastrow, C. E., E. D. Houde, and L. G. Morin. 1991. Spawning, fecundity, hatch-date frequency and young- of-the-year growth of bay anchovy Anchoa mitchilli in mid- Chesapeake Bay. Mar. Ecol. Prog. Ser. 73:161-171. 78 Abstract— Increasing interest in the use of stock enhancement as a man- agement tool necessitates a better understanding of the relative costs and benefits of alternative release strate- gies. We present a relatively simple model coupling ecology and economic costs to make inferences about optimal release scenarios for summer flounder (Paralichthys dentatus), a subject of stock enhancement interest in North Carolina. The model, parameterized from mark-recapture experiments, predicts optimal release scenarios from both survival and economic standpoints for varyious dates-of-release, sizes-at- release, and numbers of fish released. Although most stock enhancement efforts involve the release of relatively small fish, the model suggests that optimal results (maximum survival and minimum costs) will be obtained when relatively large fish (75-80 mm total length! are released early in the nursery season (April). We investigated the sensitivity of model predictions to violations of the assumption of den- sity-independent mortality by includ- ing density-mortality relationships based on weak and strong type-2 and type-3 predator functional responses (resulting in depensatory mortality at elevated densities). Depending on postrelease density, density-mortality relationships included in the model con- siderably affect predicted postrelease survival and economic costs associated with enhancement efforts, but do not alter the release scenario (i.e. combina- tion of release variables ) that produces optimal results. Predicted (from model output) declines in flounder over time most closely match declines observed in replicate field sites when mortality in the model is density-independent or governed by a weak type-3 func- tional response. The model provides an example of a relatively easy-to-develop predictive tool with which to make inferences about the ecological and economic potential of stock enhance- ment of summer flounder and provides a template for model creation for addi- tional species that are subjects of stock enhancement interest, but for which limited empirical data exist. Manuscript approved for publication 17 July 2003 by Scientific Editor. Manuscript received 20 October 2003 at NMFS Scientific Publications Office. Fish. Bull. 102:78-93 (2004). Coupling ecology and economy: modeling optimal release scenarios for summer flounder (Paralichthys dentatus) stock enhancement G. Todd Kellison David B. Eggleston Department ol Marine, Earth, and Atmospheric Sciences, North Carolina State University Raleigh, North Carolina 27695-8208 Present address (for G T. Kellison, contact author): National Park Service/ Biscayne National Park 9700 SW 328 th St, Homestead, Florida 33033 E-mail address (for G T Kellison) todd_kellison 5 nps gov Commercially important marine fish and invertebrate populations are declining worldwide in response to overexploitation and habitat degrada- tion (Rosenberg et al„ 1993; FAO 1998). This reduction in harvestable fishery resources has stimulated increasing interest in the use of hatchery-reared (HR) animals to enhance wild stocks (Munro and Bell, 1997; Travis et al., 1998; Cowx, 1999; Kent and Draw- bridge, 1999). Unfortunately, many stock enhancement programs proceed before ecological concerns are adequately addressed (Blankenship and Leber, 1996), and without the identification of goals or the evaluation of the success of enhancement efforts (Cowx, 1999). If fishery managers can satisfactorily determine that enhancement efforts will have no ecologically significant negative ramifications, then managers should establish specific, quantifiable goals and objectives of enhancement efforts as part of a responsible approach to stock enhancement (Blankenship and Leber, 1996; Heppell and Crowder, 1998). Once such goals have been established, managers should identify stocking approaches that will lead to the most cost-efficient realization of enhancement goals — a process that can be accomplished with the aid of coupled ecological and economic models. Although numerous (advanced) models (conceptual and species-specific) exist to predict the biological and ecological impact of alternative enhancement scenarios (e.g. Botsford and Hobbs, 1984; Salvanes et al„ 1992; Barbeau and Caswell, 1999; Sutton et al., 2000), there are few models ( of which we are aware) that have attempted to link the biological and ecological results of stock- ing efforts (e.g. addition of biomass to a stocked population) with the economic costs associated with various release scenarios (e.g. Botsford and Hobbs, 1984; Hobbs et al., 1990; Hernandez-Llamas, 1997; Kent and Drawbridge, 1999). Such a link is critical to the responsible use of funding to rebuild or manage fisher- ies, and for the comparison of predicted costs of enhancement versus alternative management techniques. In North Carolina, there has been recent interest in stock enhancement with summer flounder (Paralichthys dentatus) (Waters, 1996; Rickards, 1998; Waters and Mosher, 1999; Burke et al., 2000; Copeland et al. ' ) because of a combination of heavy commercial and recreational exploitation, established techniques for mass hatchery-rearing (Burke et al., 1999), and considerable knowledge of summer flounder life his- tory (Powell and Schwartz, 1977; Burke et al., 1991; Burke, 1995). Nevertheless, there have been no large-scale release experiments ( and subsequent collection of data) by which to make empirical inferences about stock enhancement potential for this species. We present a compartmental model, parameterized from mark-recapture field experiments, Copeland, B. J., J. M. Miller, and E. B. Waters. 1998. The potential for flounder and red drum stock enhancement in North Carolina. Summary of workshop, 30-31 March. 1998, 22 p. ' (Available from North Carolina State Univ, Raleigh. NC 27695.] Kellison and Eggleston: Modeling release scenarios for Paralichthys dentatus 79 Table 1 Range of numbers of summer flounder (Paralichthys dentatus) released (and resulting postrelease densities), sizes-at-release, and dates of release simulated in the model. Number released Postrelease density Size-at-release Dates of release 100-400,000 0.001-4.0 30-80 mm 1 April-15 July that incorporates size of fish released, date-of-release, and number offish released to calculate 1) predicted numbers of survivors and 2 ) economic costs associated with varying re- lease scenarios under density-independent mortality. We in- vestigated the sensitivity of model predictions to violations of the assumption of density-independent mortality because there is abundant evidence that mortality rates, or processes underlying mortality rates (e.g. growth), are affected by den- sity-dependent relationships in the wild ( see, for recent ex- amples. Bucket et al., 1999; Bystroem and Garcia-Berthou. 1999; Jenkins et al, 1999; Kimmerer et al., 2000). We did so by repeating model simulations under varying density- mortality relationships (depensatory in nature at elevated densities ), using experimental evidence from our own field studies and published observations for similar species to parameterize density-mortality relationships. Additionally, we used a scenario in which the density-mortality relation- ship changed over time to make inferences about the effect of more complex density-mortality relationships on postrelease mortality of juvenile summer flounder. Finally, we generated predicted temporal patterns of field densities under vary- ing density-mortality relationships and compared them with observed (in the field) patterns to determine whether model output under the considered density-mortality relationships matched actual patterns in the field. The model provides an example of a relatively easy-to-develop predictive tool with which to make inferences about the ecological and economic potential of stock enhancement with summer flounder and provides a template for model creation for additional species that are subjects of stock enhancement interest, but for which limited empirical data exist. Materials and methods Background In North Carolina, wild summer flounder recruit to shal- low-water estuarine nursery habitats from February to May, after which small juvenile (20-35 mm total length [TL] ) densities range from -0.1 to 1.0 fish/m 2 (Burke et al., 1991; Kellison and Taylor 2 ). Juveniles subsequently make an ontogenetic habitat shift to deeper waters ( Powell and Schwartz, 1977), apparently after reaching a total length 2 Kellison, G. T., and J. C. Taylor. 2000. Unpubl. data. De- partment of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695-8208. of -80 mm (Kellison and Taylor 2 ). By mid-July, densities of juvenile summer flounder in the shallow water nursery habitats are near zero (Kellison and Taylor 2 ). Model pathway Our compartmental model simulated the daily mortality and growth of different-size hatchery-reared (HR) fish released in the field over a 105-day period ( 1 April to 15 July, based on observed field abundances) in a hypotheti- cal release habitat of 10 hectares. The model predicted the percentage of released fish surviving and economic cost- per-survivor under 2730 release scenarios for a specified number offish released (see below). To begin the model, a value of number offish released (NFR) ranging from 100 to 400,000 (Table 1) was chosen (Fig. 1), resulting in postre- lease densities (assuming even postrelease distribution) of 0.001-4.0 fish/m 2 . These values included a range of densi- ties of juvenile summer flounder observed in wild nursery habitats ( -0-1 fish/m 2 ; mean -0.05 fish/m 2 ; Kellison and Taylor 2 ), but also included unusually high densities (>1 fish/m 2 ) in order to examine how such release strategies would affect model output (we did not examine densities >4 fish/m 2 because of a lack of data on fish response to resource limitation likely to occur as densities increased past values for which we had empirical growth data). Each group of NFR was initially assigned a "size-(TL) at-release" of 30 mm (the smallest size-at-release simulated in the model), after which a size-dependent economic cost associ- ated with the release of the 30-mm-TL fish was calculated (see below). The release group was then assigned a mini- mum Julian "day of release" of 92 (corresponding to 1 April, the earliest release date simulated in the model). A range of Julian days of release was included in the model because field-estimated growth rates were dependent on Julian day (Kellison, 2000), and growth rates are potentially impor- tant to the determination of mortality rates (Rice et al. 1993). With this model, we then calculated daily mortality and growth (described below) in the hypothetical release habitat over the number of days at large (DAL), where DAL = 197 (the Julian day corresponding to 15 July) - 92 (Julian release day), and output a number of survivors and a calculated cost- per-survivor (CPS), where CPS = cost associated with release -f predicted number of survivors, 80 Fishery Bulletin 102(1 I Input number released (NR) ' assign size-at-release (SAR) * calculate cost of release (COR) <— Size-at-release N Density- independent Julian day ' assign date of release (DOR) < I ' determine number of survivors (NOS) DAL at the beginning of the day (= «\ initial # of fish or # surviving from previous day) / / 1 da ly mortality ^ da ly growth *M * calculate number of survivors and total length (TL) at the end of the day 1 I I * output - number of survivors - cost per survivor (CPS) \ / Figure 1 Model flowchart. Dashed arrows represent model "backloops" to the indicated compartment where simulations continue with the next value of the arrow-labeled variable. Side graphs indicate the three relationships between density and mortality (number offish consumed) that were considered, and the general relationship between growth and Julian day. for the initial release scenario of fish size = 30 mm TL. Julian day = 92, and an NFR input determined by the mod- eler). The model then looped back to the "date-of-release" step and simulated the release of the 30-mm-TL fish for Julian release days 93-197, outputting a predicted number of survivors and cost-per-survivor for each release date. The model then repeated all previous steps under sequentially larger size-at-release scenarios, looping back to the "size- at-release" step and simulating the release of fish ranging in size from 32-80 mm TL fish in steps of 2 mm TL. The model output was a predicted number of survivors and economic cost-per-survivor for each release day (92-197) for each size-at-release (Fig. 1). Thus, for each input NFR, there were 26 size-at-release possibilities x 105 Julian days of release possibilities, which resulted in 2730 simulations, each of which resulted in a predicted number of survivors and cost-per-survivor for that particular release scenario. For each input NFR, the results from the 2730 simulations were plotted on two response surfaces, with an .v-axis of size-at-release, a y-axis of date-of-release, and a 2-axis of either 1) predicted number of survivors (NOS), or 2) cost- per-survivor ( CPS ), to identify release scenarios resulting in the maximum predicted number of survivors and minimum cost-per-survivor, respectively. The scenarios resulting in the maximum predicted number of survivors and minimum cost-per-survivor were not necessarily identical. Calculation of mortality, growth, survival, and economic costs associated with release During each day at large (DAL), released fish were sub- jected to a density-independent daily mortality rate of 0.02153, derived from postrelease mark-recapture data of HR summer flounder (Kellison et al., 2003b). In deriv- ing this value, mean postrelease densities were used to estimate a total number of survivors from experimental releases. Daily survival was then calculated with the equation Kellison and Eggleston: Modeling release scenarios for Paraltchthys dentatus 81 NFR x S D DAL = NOS, where NFR = number released; S D = daily survival; DAL - days at large (from release date until Julian day 197); and NOS = estimated number of survivors. Daily mortality (M D ) was then calculated from the equation M r 1-Sr At the end of each simulated day, all fish that were alive increased in growth according to the equation G D = -0.0061 x Julian day + 1.2487, which was derived from mark-recapture data (Kellison, 2000), and in which G D is daily growth in millimeters. Fish reaching 80 mm TL during the model (i.e. by 15 July) were considered to make an ontogenetic hab- itat shift to deeper waters. These fish were then subjected to one half year of natural mortality to simulate mortality- related losses from deeper-water habitats (M=0.28; Froese and Pauley, 2001). Remaining fish, now having survived -one year of natural mortality, were considered to be sur- vivors (available to the commercial fishery), which is a con- servative assumption because 1-yr-old summer flounder are only partially recruited to the commercial fishery. All fish not reaching a total length of 80 mm were assumed to perish. To determine size-dependent economic costs offish pro- duction, we used the following regression equation derived for Japanese flounder (Paralichthys olivaceus) by Sproul and Tominaga ( 1992 ) because equivalent economic data for summer flounder were unavailable: C PF = 14.24 + 1.234 x TL, where C PF = the cost per fish in Japanese yen (¥); and TL = the total length of the HR fish. Costs were then converted into US$ by using an exchange rate of 106. 7¥ per 1 US$ (universal currency converter). We feel use of this cost-of-fish-production equation is appro- priate because the Japanese flounder is closely related and similar in life history traits to the summer flounder (Tanakaet al., 1989; Burke etal., 1991 ), resulting in similar optimal rearing practices for hatchery-reared Japanese and summer flounder (Burke et al., 1999), and thus likely simi- lar rearing costs. Additionally, the scale of Japanese floun- der hatchery production is similar to, or greater than, other government subsidized hatchery production programs (e.g. red drum in Texas, cod in Norway [Svasand, 1998] ). Density-mortality relationships We tested the sensitivity of the model results (optimal predicted number of survivors and cost-per-survivor esti- mates under varying NFRs) to violations of the assumption 0.50 -i ~ 0.40 * Type 2 - weak k * Type 2 - strong E 0.30 - * ^—^-^ d Type 3 - weak ra k f ^^^^ Type 3 - strong o t 0.20 o Q. O *# ^^^^ o- 0.10 -j |^»W °°OOnnn„ 12 3 4 Density (number of fish/m 2 ) Figure 2 Proportional mortality curves for juvenile summer flounder corre- sponding to weak and strong type-2 and type-3 mortality responses. of density-independent mortality by incorporating varying types and strengths of density-dependent mortality (depen- satory in nature at elevated densities; see below) into the model. As a basis for these sensitivity analyses, we assumed that predation was the driving mechanism underlying the postrelease mortality of HR summer flounder under the densities examined (Kellison et al., 2000; Kellison et al., 2003b). Thus, we made daily mortality rates correspond to either a type-2 or type-3 predator functional response (Holling, 1959; see Lindholm et al., 2001 for example), in which proportional mortality due to predation decreases with increasing density (type-2 response) or increases ini- tially with increasing density, reaches a zenith, and then decreases with increasing density (type-3 response) (Fig. 2). Both type-2 and type-3 responses result in decreasing (depensatory) mortality at elevated prey densities due to predator satiation. We did not include scenarios in which mortality increased at elevated densities (as would be expected when densities reached those likely to result in resource limitation ) because we did not include in the model elevated release densities likely to result in resource limita- tion. We parameterized the daily mortality curves so that each response (type 2 or 3) incorporated the daily mortality rate of 0.02153. These mortality curves contain mortality values that are within ranges reported in the literature for other species of juvenile marine fishes (Bax, 1983; Houde, 1987; Nash, 1998; Rose et al, 1999). To make further infer- ences about the importance of density-dependent mortal- ity to model results, we included a 1) weak and 2) strong form of each functional response (types 2 and 3) (Fig. 2), as well as scenarios in which the response shifted temporally from 3) type 2 to 3, and 4) type 3 to 2 at the midpoint of the nursery season (Julian day 145). We included both the weak and strong forms of the type-2 and type-3 functional responses to determine the extent to which variation in the strength of the functional response would affect model pre- dictions. The strength of the functional response could vary because of annual variation in the presence or abundance of prey or because predators could affect the density-mor- tality relationship (see, for example, Hansen et al., 1998). 82 Fishery Bulletin 102(1) For example, a strong positive (compensatory) density- mortality relationship driven by predators might become weaker in years when predator abundance was lower than average. We included the temporally shifting functional response scenarios to determine the extent to which tem- poral variation in the form of the functional response would affect model predictions. Temporal variation in the form of the functional response might occur because of temporal changes in the predator community, or because of changing predator-prey size dynamics (e.g. Stoner, 1980; Black and Hairston, 1988). For example, as the nursery season for summer flounder progresses, proportionately greater num- bers of juveniles grow to sizes at which they are capable of preying on smaller juveniles (Kellison, personal obs. ). If cannibalistic summer flounder exhibit a different predatory functional response from that of the predator guild commu- nity predominating earlier in the season, then the density- mortality relationship may change seasonally. We replicated all model simulations over each of the six density-mortality relationships (weak and strong types 2 and 3, and shifting patterns [type 2 to 3 and type 3 to 2] ) to determine optimal release scenarios (maximum num- ber of survivors, minimum cost-per-survivor) under each relationship. We then compared results to those obtained under density-independent mortality to make inferences about the importance of density-mortality relationships to model results. Correspondence between predicted and observed temporal abundance patterns Different density-mortality relationships may result in distinct temporal patterns of abundance (e.g. rapid versus more gradual declines in abundance) depending on initial densities. We generated predicted patterns of temporal field abundance of juvenile summer flounder under den- sity-independent mortality and four additional density- mortality relationships (governed by weak and strong type 2 and 3 functional responses) and under varying initial densities (0.1, 0.3, and 0.5 fish/m 2 ) to examine whether the different density-mortality relationships would result in distinct temporal patterns of abundance. We used 1998-99 field data and logarithmic or polynomial regression models to generate curves that best fitted (based on r 2 values) observed (from natural nursery sites) temporal declines in abundance under varying initial densities. We compared the best-fit curves to those predicted by the model under density-independent and four additional density-mortal- ity relationships. These comparisons allowed us to make qualitative inferences about which density-mortality relationship* s) resulted in the best match between pre- dicted and observed temporal patterns of abundance. Model assumptions The assumptions of the model are the following: 1 Daily mortality is independent of size. Although there is strong evidence that mortality of fishes in the wild is size-dependent (Lorenzen, 2000 ), particularly in regard to the importance of size to susceptibility to predation (see, for example, Elis and Gibson, 1995; Furuta, 1999; Manderson et al., 1999), we found no evidence (from recaptures of released hatchery-reared fish ) of size- selective daily mortality for juvenile summer flounder ranging in size from -30-80 mm TL in shallow-water nursery areas (Kellison et al., 2003a). Implications for violations of this assumption are addressed in the "Dis- cussion" section. 2 Daily growth is independent of fish density. We based this assumption on field experiments that indicated no growth limitation at densities roughly equal to the maximum densities explored in the model (Kellison et al., 2003b). Similar findings (i.e. no food-limitation or density-dependent growth) have been reported for similar-size plaice in shallow-water nursery habitats (van der Veer and Witte, 1993). 3 Economic cost per fish (C PF ) is independent of the number of fish acquired for release (i.e. within the range of numbers offish released in model simulations, there is no decrease in cost per fish as the number of fish acquired from the production hatchery for release increases). This assumption is likely to be valid over changes in numbers of fish released common to stock enhancement programs (Sproul and Tominaga, 1992) but may not be valid as numbers released change over orders of magnitude because of economy of scale (Adams and Pomeroy 1991; Garcia et al., 1999). 4 There is no emigration from the release habitat until fish exhibit an ontogenetic shift in habitat at 80 mm TL. Although pre-ontogenetic habitat shift emigration may not truly be zero, we feel that it is also unlikely that pre- ontogenetic habitat-shift emigration accounts for more than a minimal amount of loss of released fish from the habitat of release, as supported by several points. First, rates of pre-ontogenetic shift emigration in wild juveniles are apparently low (Kellison and Taylor 2 ), suggesting that large-scale spatial migrations may not be part of the behavioral repertoire of early juvenile summer flounder. Second, irregular temporally repli- cated sampling outside of experimental release sites resulted in zero captures of emigrating hatchery-reared fish (Kellison et al., 2003b). Third, emigration rates of closely related HR Japanese flounder {Paralichthys olivaceus) are reported to be very low (Tominaga and Watanabe, 1998). In combination, these points suggest that our zero emigration assumption is appropriate. 5 Fish that do not grow to 80 mm TL during the model period (i.e. by 15 July) do not survive. Although this assumption cannot be examined with our field data, data do show that juvenile summer flounder are absent from shallow-water nursery habitats by mid to late July (Kellison et al. 3 ). Thus, all fish have either perished or made ontogenetic habitat shifts to deeper habitats by this time. Our field observations suggest that the deeper habitats to which larger flounder :t Kellison, G. T., J. C. Taylor, and J. S. Burke. 2000. Unpubl. data. Department of Marine, Earth, and Atmospheric Sciences, North Carolina State Univ., Raleigh, NC 27695-8208. Kellison and Eggleston: Modeling release scenarios for Paralichthys dentatus 83 make ontogenetic habitat shifts are inhabited by relatively high densities of potential predators (e.g. blue crabs, age 1+ flounders, red drum [Sciaenops ocellatus], searobin [Prionotus sp.], and lizardfish [Synodus sp.] ), which may be considerably less abundant in shallow-water habitats. These relatively large and abundant predators would presumably expose small migrating fish to high rates of predation (see, for example, Elis and Gibson, 1995; Furuta, 1999; Manderson et al„ 1999). This assumption is supported by research with the congener Japanese flounder (Paralich- thys olivaceus). Although a range of sizes of hatchery-reared Japanese flounder may survive within relatively shallow nursery habitats, fishes less than 90 mm TL moving into relatively deep waters are poorly rep- resented in subsequent age classes, most likely due to predation-induced mortality (Yamashita et al., 1994; Furuta, 1999). There is no relationship between length of rearing period (time spent in the hatchery environment) and probability of postrelease mortality related to behavioral deficits (Olla et al., 1998). Hatchery-specific selection pressures may result in HR fish that are behaviorally selected to survive in the hatchery and not in the wild (see Olla et al., 1998; Kellison et al., 2000; for discus- sion). We assume that behavioral deficits are not exacerbated with time spent in the hatchery (i.e. behavioral deficits are equal for all sizes-at-release). Results The most important factor affecting the number of survivors (and therefore percent survival) was size-at-release because the greatest numbers and percentages of survi- vors were always produced by releasing the largest fish possible (80 mm TL in the model). Number of survivors decreased with decreas- ing size-at-release and with increasing Julian day of release (Fig. 3A). The cost-per-survivor ( CPS ) was also most affected by size-at-release, such that CPS decreased with increasing size- at-release (Fig. 3B). CPS generally increased with increasing Julian day of release (Fig. 3B), although this effect was less important than the effect of size-at- release. Because mortality was originally assumed to be density-independent, the optimal cost-per-survivor did not vary with the number offish released (Fig. 4), and the relationship between number offish released and number of survivors was linear (Fig. 4), such that the maximum number of survivors were generated from the greatest number offish released (NFR=400,000). 220 20 80 90 220 Figure 3 Response surfaces of iAi number offish survivors (summer flounder I and (Bi cost-per-survivor (CPS) as a function of date of release and size at release at number released (NR) = 5000 (postrelease density=0.05) under density-independent mortality. CPS values greater than $10 were set equal to $10 for ease of presentation. Sensitivity of model predictions to violations of density-independent mortality assumption Model results varied considerably under the various den- sity-mortality relationships (Fig. 5, A and B), indicating the importance of knowledge of the relationship between numbers of fish released (density) and mortality in the wild to predicting optimal release scenarios. Variation in model output was dependent on the type and strength of Fishery Bulletin 102(1) the density-mortality relationship. For example, at postre- lease densities of 0.5 fish/m 2 (NFR=50,000), survival of released flounder under density-independent mortality was ~28% higher than that predicted under strong type-3 mortality, but only -2% higher than that predicted under weak type-2 mortality (Fig. 5A). At postrelease densities of 0.001 fish/m 2 (NFR=100), survival of released flounder under density-independent mortality was ~41% higher 450000 -I m 400000 • § 350000 • £ 300000 • « 250000 ; ° 200000 E 150000 | 100000 z 50000 — — optimal number of survivors : — o— optimal CPS ^^^" r 1 60 : 1 50 O 1.40 g 1 30 -g -1,20 5 [110 <§ - 1 00 g ■0.90 < ■0 80 - 0.70 C 0-60 W 50000 10000 15000 20000 25000 30000 35000 40000 Number released Figure 4 Optimal number of fish survivors and cost-per-survivor as a function of varying numbers of summer flounder released under density-indepen- dent mortality. 12 3 4 5 Density (number of fish/m 2 ) Figure 5 l A i Optimal percent survival and iBi optimal cost-per-survival (US$) as a func- tion of postrelease density undci density-independent and varying density- dependent, mortality relationships for summer flounder. than that predicted under strong type-2 mortality, but -2% less than that predicted under strong type-3 mortality ( Fig. 5A). In contrast, when postrelease densities were relatively high, there was less of an impact of density-mortality rela- tionship on postrelease survival and costs associated with stock enhancement. For example, at postrelease densities of three fish/m 2 (NFR=300,000), survival of released floun- der differed by less than 4% between density-independent, weak or strong type-2, and weak type-3 mor- tality, although survival under strong type-3 mortality was ~99c less than that predicted under density-independent mortality and -11% less than that predicted under strong type-2 mortality (Fig. 5A). Thus, the model results were most sensitive to violations of the assumption of density-independent mortality at low densities offish released in the field. Type-2 mortality As with density-indepen- dent mortality, the most important factor affecting number of survivors and cost per survivor under type-2 mortality was size-at- release (Fig. 6, A and B). In all simulations, the greatest number of survivors was pro- duced by releasing the largest fish possible. Number of survivors decreased with increas- ing Julian day of release (Fig. 6A). There was a considerable interaction between size- at-release and number of fish released, such that low postrelease densities were subjected to relatively high proportional mortality. Thus, when fish were released in low numbers and at small sizes, the fish were subjected to relatively high proportional mortality rates for long periods of time (while they grew towards the 80-mm-TL ontogenetic shift size) and consequently produced few or no survi- vors (Fig. 6A). Optimal release scenarios under strong type-2 mortality produced substantially lower (>40% in some cases) percent survival (and therefore substantially higher cost-per-survivor) estimates at low to moderate numbers released (NFR= 100-50,000; postrelease density=0.001-0.5 fish/m 2 ) than under density-independent mortality (Fig. 5, A and B). Differences in percent survival estimates (and thus cost-per-survivor estimates) between density-indepen- dent survival and weak or strong type-2 mortality declined to less than 5 r i when the numbers released increased to 25,000 (postrelease density=0.25 fish/m 2 ) under weak type-2 mortality and 75.000 (postrelease density=0.75 fish/m 2 ) under strong type-2 mortality (Fig. 5A). Thus, model predictions under density-inde- pendent mortality differed most from predictions under mortality governed by - density-independent -type 2 - weak - type 2 - strong -type 3 - weak ■type 3 - strong density-independent type 2 - weak type 2 - strong type 3 - weak type 3 - strong Kellison and Eggleston: Modeling release scenarios for Paralichthys dentatus 85 B a type-2 predator functional response when postrelease densities were relatively low. Type-3 mortality As in all other simulations, the most important factor affecting number of survivors under type-3 mortality was size- at-release, such that the greatest numbers of survivors were always produced by releasing the largest fish possible (Fig. 7A). Number of survivors decreased with increasing Julian day of release (Fig. 7A). Percent survival was considerably lower (>25% in some cases) under type-3 mortality than under density- independent mortality at moderate to high numbers released (NFR=10, 000-400, 000) (Fig. 5 A). In nearly all simulations, the lowest CPS values were produced by releasing the larg- est fish possible (Fig. 7B). The exceptions to the "large size = optimal CPS" rule occurred when postrelease densities were small (cor- responding to numbers released of 100, 500, and 1000) and the mortality curve was type 3 (weak or strong). In these instances, mortality was sufficiently low at low release densities ( Fig. 7B ) so that the difference in overall sur- vival between small- and large-released fish was small enough to be overridden by the in- creased cost of the larger fish, and the mini- mum CPS was obtained when small (42-44 mm TL) fish were released (e.g. Fig. 7B). At low numbers released (NFR=100-1000), optimal cost-per-survivor was considerably lower (>45% in some cases) under type-3 mortality than under density-independent mortality (Fig. 5A). As NFR increased, CPS under type-3 mortality became greater ( -40^ in some cases) than that achieved under den- sity-independent mortality (Fig. 5B). Temporal shift in functional response from type 2 to type 3, and from type 3 to type 2 The optimal numbers of survivors under varying numbers released were identical, and optimal CPS values nearly identical, when the form of the functional response changed from a type 2 to a type 3, and from a type 3 to a type 2, midway through the juvenile nurs- ery season (Fig. 8, A and B). The differences at low postrelease densities between optimal CPS values under shifting type 2 to type 3 and type 3 to type 2 scenarios (Fig. 8A) occurred because initial mortality under the type-3 functional response was sufficiently low that the difference in overall survival between small- and large-released fish was small enough to be overridden by the increased cost of the larger fish (Fig. 8A). The minimum CPS was obtained when small (42-44 mm TL) fish were released (in all other cases, optimal results were obtained when size-at-release was maximized) (Fig. 8A). The major difference between the two shifting scenarios is that the re/ease Figure 6 Response surfaces of (A) number offish (summer flounder I survivors and (B) cost-per-survivor (CPS) as a function of date of release and size at release at number released (NR) = 5000 (postrelease density=0.05l under a strong type-2 functional response. CPS values greater than $10 were set equal to $10 for ease of presentation. release dates producing optimal results for a given number of fish released varied depending on the direction of the shifting functional response. For example, when the func- tional response shifted from a type 2 to a type 3, a release of 100,000 HR organisms achieved optimal results when release occurred early in the season (Julian day <145) (Fig. 9A). When the functional response shifted from a type 3 to a type 2, a release of 100,000 HR summer floun- der achieved optimal results only when releases occurred later in the season (Julian day >145) (Fig. 9B). When the 86 Fishery Bulletin 102(1) functional response shifted from a type 3 to a type 2, releas- ing 100,000 HR organisms prior to Julian day 146 resulted in markedly decreased survival (and therefore increased CPS ) compared to results obtained from releases after day 146 (e.g. releasing on Julian day 92 resulted in a decrease in number of survivors and an increase in CPS of 22.8% and 29.7%, respectively) (Fig. 9B). Thus, date-of-release had a significant effect on the results (and therefore in determining optimal release strategies) when the relation- ship between density and mortality changed temporally, suggesting that the presence of a temporal shift in the func- 500 £ 400 300 200 E z 100 220 OaV' Size at re/ease Figure 7 Response surfaces of (A) number offish (summer flounder) survivors and (B) cost-per-survivor (CPS) as a function of date of release and size at release at number released (NR) = 500 (postrelease density=0.005) under a strong type-3 functional response. CPS values greater than $10 were set equal to $10 for ease of presentation. tional response of the predator guild would have consider- able effects on the number of survivors and CPS for stock enhancement efforts with juvenile summer flounder. Correspondence between predicted and observed temporal abundance patterns Under the assumption of a type-2 functional response, predicted declines in juvenile summer flounder density over time were rapid when initial density was relatively low (i.e. 0.1 fish/m 2 ) (Fig. 10, A and B). These predictions contrast with those observed in the field, in which declines at relatively low initial densities were gradual (compare Fig. 10A and 10B to Fig. 10F). Under the assumption of a type-3 functional response, predicted declines were rapid when initial density was relatively high (i.e. 0.5 fish/m 2 ) I Fig. 10, C and D). These results generally contrast with those observed in the field, in which declines at relatively high densities were much less rapid than those predicted under a strong type-3 functional response, and somewhat less rapid than those predicted under a weak type-3 functional response (Figs. 10F and 11). Under density-independent mortality, there was little difference in predicted declines in juvenile summer flounder density over time between the three initial density levels (0.1, 0.3, and 0.5 fish/m 2 ); in each case there was a gradual decrease in density over time (Fig. 10E). These results were similar to those observed in the field, although declines at rel- atively high densities in the field were some- what more rapid than those predicted under density-independent mortality ( compare Figs. 10E and 10F). Thus, a density-mortality rela- tionship lying between that generated under density-independence and that generated under the weak type-3 functional response in the model would most closely predict the temporal declines observed in the field. Discussion Implications for stock enhancement of summer flounder Regardless of the relationship between den- sity and mortality, size-at-release was the most important variable in the model affect- ing survival and costs associated with stock enhancement of summer flounder. The model predicts that under all release scenarios, 1) survival will be maximized and 2) costs asso- ciated with stock enhancement (i.e. cost per survivor) will be minimized when HR fish are released at the largest size possible. From a survival standpoint, these results are not Kellison and Eggleston: Modeling release scenarios for Paralichthys dentatus 87 surprising. Larger fish spend fewer days than smaller fish in the wild nursery habitats before making an ontogenetic habitat shift to deeper waters and thus are susceptible to daily natural mortality for fewer numbers of days than are smaller fish. Thus, total mortality of smaller fish is greater than that of larger fish. Additionally, although we chose to make mortality independent of size in the model, abundant literature suggests that natural mortality (especially due to predation ) may decrease with increasing size by mecha- nisms such as enhanced resistance to starvation, decreased vulnerability to predators, and better tolerance of environ- mental extremes (Sogard, 1997; Hurst and Conover, 1998; Lorenzen, 2000). Thus, the difference in predicted survival between 1 ) relatively large and relatively small fish and 2 ) fish released early versus late in the season in our model would be even greater if larger summer flounder suffered lower natural mortality than smaller fish. Furthermore, the daily mortality estimate used in the density-inde- pendent simulations and to parameterize the different types of density-mortality relationships may have been an underestimate of daily mortality (Kellison, 2000). If a greater estimate of daily mortality had been used, the dif- ference in predicted survival between relatively large and relatively small fish in our model would have been further exacerbated because smaller fish spend longer amounts of time in the model growing to the 80-mm-TL ontogenetic shift size. These conclusions are supported by empirical research demonstrating that relatively large released HR fish suffer lower mortality than relatively small HR fish released in the field (e.g. Yamashita et al., 1994; Leber, 1995; Willis et al., 1995; Tominaga and Watanabe, 1998; Svasandetal.,2000). Although the survival predictions of the model (total mortality decreases with increasing size-at-release) are not surprising, the economic (cost-per-survivor) predic- tions were unexpected. The paradigm for stock enhance- ment strategy is that the rearing of relatively large fish for release is cost prohibitive, so that mass releases of relatively small, inexpensive-to-rear fish are a better strategy than the release of larger, expensive-to-rear fish (Kellison, personal obs.). Thus, relatively small juveniles are released in virtually all current stock enhancement programs (e.g. Russell and Rimmer, 1997; Masuda and Tsukamoto, 1998; McEachron et al., 1998; Svasand, 1998; Serafy et al., 1999). Nevertheless, large-scale hatcheries and grow-out facilities are using ever-increasing technol- ogy to minimize the costs associated with the production of relatively large fishes (Sproul and Tominaga, 1992). Thus, for species for which 1) hatcheries are capable of producing relatively large fish at relatively low costs (as is likely for summer flounder), and 2) postrelease survival rates increase with release size, release scenarios utilizing the largest fish possible may maximize the potential (i.e. produce maximum survival at minimum costs ) of stock en- hancement efforts. In these cases, the "small fish maximize stock enhancement potential" paradigm might be replaced with a "large fish maximize potential" paradigm. As a ca- veat, this "large fish" strategy may be limited by spatial limitations of hatcheries in producing large numbers of relatively large fish. Because reared fish generally must 1 40 i */» ^_ 1.20- o 2 1 00- w 80- (1> Q. 60- If) O 40- O Type 2 to 3 Type 3 lo 2 20-1— -! , 1 r Postrelease density Figure 8 Optimal lA) economic cost-per-survivor and (B) per- cent survival of released hatchery-reared summer flounder under temporally shifting functional re- sponses of type 2 to type 3 and type 3 to type 2. be kept below critical densities in hatchery environments because of water quality and fish interaction issues (e.g. cannibalism), larger fish necessarily require more space than smaller fish for rearing. If the demand for space to rear large quantities of large fish can be realized, then the model simulations indicate that stock enhancement strat- egies in which size-at-release is maximized will produce the maximum number of survivors. Although not as important as size-at-release, Julian day of release had a significant effect on survival and cost-per- survivor in the model, such that enhancement efforts were always more successful (more survivors, lower costs) when fish were released at the earliest Julian day possible. These results occurred because growth in the model decreased with increasing Julian Day. Although the mechanisms un- derlying this decrease in growth with increasing Julian day are unknown, they may be related to decreased prey avail- ability or metabolic efficiency as temperatures increase with increasing Julian day (Malloy and Targett, 1994a, 1994b; Fujii and Noguchi, 1996; Howson, 2000). Thus, for a given size-at-release, fish released earlier in the season experienced greater growth rates than fish of the same size-at-release released later in the season and therefore reached the 80-mm-TL ontogenetic shift size faster (over a period of fewer days) than fish released later in the season. Thus, fish released earlier in the season were susceptible to natural mortality for fewer days than fish released later in the season and therefore suffered lower total mortality. These results emphasize the importance of knowledge of possible time-dependent growth in the field prior to stock enhancement efforts. Fishery Bulletin 102(1) Is density important? Effects of varying density-mortality relationships Our results suggest that the relationship between density and mortality has the potential to significantly affect opti- mal release scenarios associated with stock enhancement efforts. Because the original simulations were performed under density-independent mortality, the number of survivors originally increased linearly with the number B 1e+5 (/> 8e+4 o > > 6e+4 tfl n m 4e+4 a h 3 2e+4 z 80 released, resulting in a density-independent cost-per- survivor. Thus, when mortality is independent of density (over a given range of densities) for a target species for stock enhancement, managers will maximize the number of survivors produced by releasing the greatest number of fish possible within that range for a given size class. When mortality varied with density of released fish, the number of survivors and cost-per-survivor depended on the den- sity-mortality relationship. In some cases, optimal results (maximum survival and minimum cost) differed depending on whether the response variable was number of survivors or cost-per-survivor. Under the assumption of a strong type-3 functional response and under relatively low postrelease densities, survival was optimized (maximized) by releasing the largest fish ( 80 mm TL) possible; however, cost-per-survivor was optimized (mini- mized) by releasing smaller fish (42-44 mm TL). This result occurred because mortality at low postrelease densities was sufficiently low that the difference in total mortality attributed to the longer "susceptibility" period of the smaller fish was insufficient to override the economic advan- tage of releasing smaller fish. Simulations under shifting functional responses (type 2 to type 3 and type 3 to type 2) produced optimal results similar to those obtained when nonshifting type- 2 or type-3 functional responses were employed because densities were generally reduced to such low numbers by the time the shift occurred that the changing density-mortality relationship was inconsequential. Importantly, when functional responses shifted temporally, the predicted number of survivors and economic cost per survivor was at times very dependent on date of release, suggesting that identifying or ruling out shifting functional responses in the wild may be critical to accurate prediction of response vari- ables (survivors and economic costs) associated with stock enhancement. Although we are not aware of reports in the literature of shifting functional responses in the wild, we are also not aware of studies that have tested for such a phenomenon, possibly because of the logisti- cal difficulties inherent in identifying a shifting functional response. Correspondence between predicted and observed temporal abundance patterns Figure 9 (A) Response surface of optimal number of summer flounder survivors as a function of date of release and size at release at number released (NR) = 100,000 (postrelease density=1.0 fish/m 2 1 under the assumption of a temporally shifting functional responses from type 2 to type 3. Response surfaces of optimal number of survivors as a function of date of release and size at release at number released (NR) = 100. 000 (postrelease density=1.0 fish/m 2 > under the assumption of a temporally shifting functional responses from type 3 to type 2. Predictions of field abundance patterns of juve- nile flounder density over time were noticeably different under density-independent mortality and density-dependent mortality governed by type-2 and type-3 functional responses. For example, our simulations predict that fish den- sity should decrease rapidly under relatively low initial densities if the functional response is type 2, decrease rapidly at relatively high initial densities if the functional response is type 3, and Kellison and Eggleston: Modeling release scenarios for Parahchthys dentatus 89 OS- 04- 0.3 0.2 01 00 E Strong type 2 Strong type 3 150 Dl B Weak type 2 05 0.4 3 02 110 130 150 170 190 210 D Weak type 3 Julian day Figure 10 Predicted temporal trends in summer flounder abundance under initial densities of 0.5, 0.3, and 0.1 fish/m 2 under the assumption of a functional response that is a (A) strong type 2. IB) weak type 2, (C) strong type 3. i D i weak type 3. and under the assumption of (E) density-independent iDIl mortality. The curves in iFi are best fitted (highest r 2 value) to data collected in Duke Beach 1999 (curve a, r 2 =0.82). Haystacks Marsh 1999 (curve b, r 2 =0.73), Prytherch Marsh 1999 (curve c. ;- 2 =0.82), Towne Beach 1999 (curve d, r 2 =0.91). Radio Beach 1999 (curve e, r 2 = 0.27), Duke Beach 1998 (curve f, r 2 =0.31), and Prytherch 1998 (curve g, r 2 =0.16) (see Fig. 11 for data). gradually decrease regardless of initial density if mortal- ity is density independent. From examinations of tempo- ral abundance patterns from several nursery sites (see Kellison et al., 2003b, for site descriptions), it is evident that observed declines at relatively low initial densities are similar to predicted declines under both density-inde- pendent mortality and a weak type-3 functional response; whereas observed declines at relatively high initial densi- ties are somewhat less gradual than predicted under den- sity-independent mortality, but somewhat more gradual than predicted under the weak type-3 functional response. These results suggest that model predictions made under the assumption of a weak type-3 response may give rea- sonably accurate but conservative predictions of juvenile summer flounder mortality and economic costs associated with stock enhancement for comparison with alternative management methods. As a caveat, although we found no evidence of size-dependent daily mortality over the range of fish sizes examined in this study, it is very likely that such a relationship exists to some extent (Sogard, 1997; Lorenzen, 2000). Incorporating size-dependent mortality into the model would decrease the slopes of the predicted temporal abundance curves but should not change the conclusion that the observed data lie somewhere between values predicted under density-independent mortality and those governed by a weak type-3 functional response, respectively. Additionally, because the portions of the curves used to delineate between temporal abundances expected under density-independent versus varying den- sity-mortality relationships are from early in the growth season (later parts of the curve converge on very low den- sities) and because nearly all fish in these portions of the curves are at sizes well below that at which ontogenetic emigration occurs, the exclusion of emigration from these simulations should not affect the general conclusions reached. These issues could be clarified with further field trials to investigate the dependence of daily mortality rates on fish size. 90 Fishery Bulletin 102(1) E E Prytherch 1999 Radio 1999 003 Prytherch 1998 * r* = 0.1575 02 001 * _. 95 105 115 125 135 145 155 165 B 03 Duke 1999 2 • r = 8162 01 9* * ♦*4U** T M T *' •*♦ D Haystacks 1999 Towne 1999 r" i 9063 ^s^ • "" "Y — » '«W. W. LT» 95 115 135 155 175 195 Duke 1998 1 * r 2 = 03113 0.05 • * **> ♦. • ~~7 • 95 115 135 155 175 195 Julian day Julian day Figure 11 Temporal density patterns from (A) Duke Beach, 1999; (B) Haystacks Marsh, 1999; (C) Prytherch Marsh, 1999; (D) Towne Beach, 1999; (E) Radio Beach, 1999; (F) Duke Beach, 1998; and (G) Prytherch Marsh 1998. Densities are corrected for gear bias (see Kellison, 2000). Model utility and implications Although model results varied considerably under the various density-mortality relationships, the overall pre- dictions that survival would be maximized and economic costs minimized when relatively large fish were released early in the season were unaffected by the density- mortality relationship. These results suggest that manag- ers may use this model to make inferences about optimal release scenarios even if density-mortality relationships are unknown. Additionally, these results have important implications for the cost efficiency of stock enhancement programs. Managers can use the model to determine the release scenarios under which they can 1) maxi- mize the number of survivors, given a financial limit (e.g. given a budget of x dollars, what release scenario or scenarios will produce the greatest number of survi- vors?), and 2) minimize costs, given a goal of number-of- survivors-produced (e.g. given a goal of producing .v survivors, what release scenario or scenarios will be most cost efficient?). In conclusion, the compartmental model used in this study provides an example of a relatively easy-to-develop predictive tool with which to make inferences about the ecological and economic potential of stock enhancement, in relation to alternative management approaches, to rebuild depleted fisheries. Kellison and Eggleston: Modeling release scenarios for Paraltchthys dentatus 91 Acknowledgments We thank Brian Burke (NCSU) for tutelage in the use of Visual Basic. Mike Denson (South Carolina Department of Natural Resources) and Pete Schuhmann (UNC-Wilm- ington ) greatly contributed to the editing of an earlier ver- sion of this manuscript. Mark Wuenschel, Michael Martin, Brian Degan, Lisa Etherington. and Mikael Currimjoe pro- vided valuable laboratory and field assistance necessary for parameter estimation. This project was partially funded by the University of North Carolina at Wilmington/North Carolina State University Cooperative Ph.D. Program, and a grant from the National Science Foundation (OCE 97- 34472) to D. Eggleston. Literature cited Adams, C. M.. and R. S. Pomeroy. 1991. Scale economies in hard clam aquaculture. J. Shell- fish Res. 10:307-308. Barbeau. M. A., and H. Caswell. 1999. A matrix model for short-term dynamics of seeded populations of sea scallops. Ecol. Appl. 9:266-287. Bax, N. J. 1983. Early marine mortality of marked juvenile chum salmon iOncorhynchus keta) released into Hood Canal, Puget Sound, Washington, in 1980. Can. J. Fish. Aquat. Sci. 40:426-435. Black, R. W. II, and N. G. Hairston Jr. 1988. Predator driven changes in community structure. Oecologia 77:468-479. Blankenship, H. L., and K. M. Leber. 1996. A responsible approach to marine stock enhancement. In Developing and sustaining world fish- eries resources: the state of science and management (D. A. Hancock, D. C. Smith, A. Grant, and J. P. Beumer, eds. I, p. 485^91. CSIRO, Collingwood, Australia. Botsford, L. W„ and R. C. Hobbs. 1984. Optimal fishery policy with artificial enhancement through stocking: California's white sturgeon as an example. Ecol. Model. 23:293-312. Buckel, J. A., D. O. Conover, N. D. Steinberg, and K. A. McKown. 1999. Impact of age-0 bluefish iPomatomus saltatrix) preda- tion on age-0 fishes in the Hudson River estuary: evidence for density-dependent loss of juvenile striped bass iMorone saxatilis). Can. J. Fish. Aquat. Sci. 56:275-287. Burke, J. S. 1995. Role of feeding and prey distribution on summer and southern flounder in selection of estuarine nursery habitats. J. Fish Biol. 47:355-366 Burke. J. S., J. M. Miller, and D. E. Hoss. 1991. Immigration and settlement pattern of Paralichthys dentatus and P. lethostigma in an estuarine nursery ground. North Carolina, U.S.A. Neth. J. Sea Res. 27:393^405. Burke, J. S., J. P. Monaghan, and S. Yokoyama. 2000. Efforts to understand stock structure of summer flounder {Paralichthys dentatus) in North Carolina, USA. J. Sea Res. 44:111-122. Burke, J. S., T. Seikai, Y. Tanaka, and M. Tanaka. 1999. Experimental intensive culture of summer flounder, Paralichthys dentatus. Aquaculture 176: 135-144. Bystroem, P., and E. Garcia-Berthou. 1999. Density dependent growth and size specific competi- tive interactions in young fish. Oikos 86: 217-232. Cowx, I. G. 1999. An appraisal of stocking strategies in the light of devel- oping country constraints. Fish. Manag. Ecol. 6: 21-34. Elis, T., and R. N. Gibson. 1995. Size-selective predation of 0-group flatfishes in a Scottish coastal nursery ground. Mar. Ecol. Prog. Ser. 127:27-37. FAO (Food and Agriculture Organization of the United Nations). 1998. Summary: The state of world fisheries and aquacul- ture. http://www.fao.org/WAICENT/FAOINFO/FISHERY/ FISHERY.HTM. [Accessed: June 1998.1 Froese, R., and D. Pauly, eds. 2001. FishBase. World wide web electronic publication. http://www.fishbase.org. [Accessed: January 2001.] Fujii, T, and M. Noguchi. 1996. Feeding and growth of Japanese flounder {Paralich- thys olivaceus) in the nursery ground. In Proceedings of an international workshop: survival strategies in early life stages of marine resources, p. 141-154. A. A. Balkema Publishers, Brookfield, VT. Furuta, S. 1999. Seasonal changes in abundance, length distribution, feeding condition and predation vulnerability of juvenile Japanese flounder, Paralichthys olivaceus, and prey mysid density in the Tottori coastal area. Nippon Suisan Gak- kaishi 65:167-174. Garcia, L. MaB.. R. F. Agbayani, M. N. Duray, G. V. Hilomen-Garcia, A. C. Emata, and C. L. Marie. 1999. Economic assessment of commercial hatchery produc- tion of milkfish (Chanos chanos Forsskal) fry. J. Applied Ichthyol. 15:70-74. Hansen, M. J., M. A. Bozek, J. R. Newby, S. P. Newman, and M. D. Staggs. 1998. Factors affecting recruitment of walleyes in Escanaba Lake, Wisconsin. 1958-1996. North Am. J. Fish. Manag. 18:764-774. Heppell, S. S., and L. B. Crowder. 1998. Prognostic evaluation of enhancement programs using population models and life history analysis. Bull. Mar. Sci. 62:495-507. Hernandez-Llamas, A. 1997. Management strategies of stocking density and length of culture period for the Catarina scallop Argopecten circu- lars (Sowerby): A bioeconomic approach. Aquacult. Res. 28:223-229. Hobbs, R. C, L. W Botsford, and R. G. Kope. 1990. Bioeconomic evaluation of the culture/stocking concept for California halibut. The California halibut. Paralichthys californicus, resource and fisheries, 1990. p. 417^150. Cal. Dep. Fish Game Fish. Bull., vol. 174. Holling, C. S. 1959. Some characteristics of simple types of predation and parasitism. Canadian Entomologist 91: 385-398. Houde, E. D. 1987. Fish early life history dynamics and recruitment variability. Am. Fish. Soc. Symp. 2:17-29. Howson, U. A. 2000. Nursery habitat quality for juvenile paralichthyid flounders: Experimental analysis of the effects of physico- chemical parameters. Ph.D. diss., 129 p. LTniv. Delaware, Newark, DE. 129 p. Hurst, T P.. and D. O. Conover. 1998. Winter mortality of young-of-the-year Hudson River bass (Morone saxatilis): Size-dependent patterns and effects on recruitment. Can. J. Fish Aquat. Sci. 55: 1122-1130. 92 Fishery Bulletin 102(1) Jenkins, T. M. Jr., S. Diehl. K. W. Kratz. and S. D. Cooper, 1999. Effects of population density on individual growth of brown trout in streams. Ecol. 80:941-956. Kellison, G. T. 2000. Evaluation of stock enhancement potential for summer flounder (Paraliehthys dentatus): an integrated laboratory, field and modeling study. Ph.D. diss., 196 p. NC State Univ., Raleigh, NC. Kellison, G. T., D. B. Eggleston, and J. S. Burke. 2000. Comparative behaviour and survival of hatch- ery-reared versus wild summer flounder {Paraliehthys dentatus). Can. J. Fish. Aquat. Sci. 57:1870-1877. Kellison, G. T., D. B. Eggleston, J. C. Taylor, and J. S. Burke. 2003a. An assessment of biases associated with caging, tethering, and habitat-specific trawl sampling of summer flounder {Paraliehthys dentatus). Estuaries 26:64-71. Kellison G. T„ D. B. Eggleston. J. C. Taylor, J. S. Burke, and J. A. Osborne. 2003b. Pilot evaluation of summer flounder stock enhance- ment potential using experimental ecology. Mar. Ecol. Prog. Ser. 250:263-278. Kent, D. B., and M. A. Drawbridge. 1999. Developing a marine ranching programme: A multi- disciplinary approach. Marine ranching: Global perspec- tives with emphasis on the Japanese experience. FAO fisheries circular 943:66-78. FAO, Rome. Kimmerer, W. J., J. H. Cowman Jr., L. W. Miller, and K. A. Rose. 2000. Analysis of an estuarine striped bass iMorone saxa- tilis) population: influence of density-dependent mortality between metamorphosis and recruitment. Can. J. Fish. Aquat. Sci. 57:478-486. Leber, K. M. 1995. Significance of fish size-at-release on enhancement of striped mullet fisheries in Hawaii. J. World. Aquacult. Soc. 26:143-153. Lindholm, J. B., P. J. Auster, M. Ruth, and L Kaufman. 2001. Modeling the effects of fishing and implications for the design of marine protected areas: juvenile fish responses to variation in seafloor habitat. Conserv. Biol. 15:424-437. Lorenzen, K. 2000. Allometry of natural mortality as a basis for assessing optimal release size in fish stocking programmes. Can. J. Fish. Aquat. Sci. 57:2374-2381. Malloy, K. D., and T. E. Targett. 1994a. The use of RNA:DNA ratios to predict growth limita- tion of juvenile summer flounder {Paraliehthys dentatus) from Delaware and North Carolina estuaries. Mar. Biol. 118:367-375. 1994b. Effects of ration limitation and low temperature on growth, biochemical condition, and survival of juve- nile summer flounder from two Atlantic Coast nurseries. Trans. Am. Fish. Soc. 123:182-193. Manderson, J. P., B. A. Phelan, A. J. Bejda, L. L. Stehlik, and A. W. Stoner. 1999. Predation by striped searobin (Prionotus evolans, Triglidae) on young-of-the-ycar winter flounder (Pscinln pleuronectes americanus, Walbaum): examining prey size selection and prey choice usinj; field observations and labo- ratory experiments. ,1. Exp. Mar. Biol. Ecol. 242:211-231. Masuda. R. and K. Tsukamoto. 1998. Stock enhancement in Japan: review and perspective. Bull. Mar. Sci. 62:337-358. McEachron, L. W.. R. L. Colura, B.W. Bumguardner, and R. Ward. 1998. Survival of stocked red drum in Texas. Bull. Mar. Sci. 62:359-368. Munro, J. L., and J. D. Bell. 1997. Enhancement of marine fisheries resources. Rev. Fish. Sci. 5:185-222. Nash, R. D. M. 1998. Exploring the population dynamics of Irish Sea plaice, Pleuronectes platessa L.. through the use of Paulik diagrams. J. Sea Res. 40:1-18. Olla, B. L., M. W. Davis, and C. H. Ryer. 1998. Understanding how the hatchery environment represses or promotes the development of behavioural sur- vival skills. Bull. Mar. Sci. 62:531-550. Powell, A. B., and F J. Schwartz. 1977. Distribution of paralichthid flounders (Bothidae: Paraliehthys) in North Carolina estuaries. Chesapeake Sci. 18: 334-339. Rice, J. A., T. J. Miller, K. A. Rose, L. B. Crowder. E. A. Marschall, A. S. Trebitz, and D. L. DeAngelis. 1993. Growth rate variation and larval survival: Inferences from an individual-based size-dependent predation model. Can. J. Fish. Aquat. Sci. 50:133-142. Rickards, W L. 1998. Sustainable flounder culture and fisheries: a regional approach involving Rhode Island, New Hampshire, Vir- ginia, North Carolina, and South Carolina. In Nutrition and technical development of Aquaculture, Nov. 1998. p. 17-20. Sea Grant, Durham, NH. Rose, K. A., J. H. Cowan Jr, M. E. Clark, E. D. Houde, and S-B Wang. 1999. An individual-based model of bay anchovy popula- tion dynamics in the mesohaline region of Chesapeake Bay. Mar. Ecol. Prog. Ser. 185:113-132. Rosenberg, A. A., M. J. Fogarty, M. P. Sissenwine, J. R. Beddington, and J. G. Shepard. 1993. Achieving sustainable use of renewable resources. Science 262:828-829. Russell, D. J., and M. A. Rimmer. 1997. Assessment of stock enhancement of barramundi Lates ealearifer (Bloch) in a coastal river system in far northern Queensland. Australia. Developing and sustaining world fisheries resources. In Developing and sustaining world fisheries resources: the state of science and management, p. 498-503. CSIRO, Collingwood (Australia I. Salvanes, A. G. V., D. L. Aksnes, and J. Giske. 1992. Ecosystem model for evaluating potential cod produc- tion in a west Norwegian fjord. Mar. Ecol. Prog. Ser. 90: 9-22. Serafy, J. E„ J. S. Ault, T. R. Capo, and D. R. Schultz. 1999. Red drum, Sciaenops ocellatus L., stock enhance- ment in Biscayne Bay. FL, USA: assessment of releasing unmarked early juveniles. Aquacult. Res. 30:737-750. Sogard, S. M. 1997. Size-selective mortality in the juvenile stage of tele- ost fishes: a review. Bull. Mar. Sci. 60: 1 129-1 1 57. Sproul. J. T. and O. Tominaga. 1992. An economic review of the Japanese flounder stock enhancement project in Ishikari Bay. Hokkaido. Bull. Mar. Sci. 50:75-88. Stoner, A.W. 1980. Feeding ecology of Lagodon rhomboides (Pisces: Spari- dae): variation and functional responses. Fish. Bull. 78: 337-352. Sutton, T. M„ K. A. Rose, and J. J. Ney. 2000. A model analysis of strategies for enhancing stocking success of landlocked striped bass populations. N. Am. J. Fish. Manag. 20:841-859. Kellison and Eggleston: Modeling release scenarios for Paralichthys dentatus 93 Svasand. T. 1998. Cod enhancement studies in Norway — background and results with emphasis on releases in the period 1983- 1990. Bull. Mar. Sci. 62:313-324. Svasand, T. S., T. Kristiansen, T. Pedersen. A. G. V Salvanes, R. Engelsen, G. Naevdal, and M. Nodtvedt. 2000. The enhancement of cod stocks. Fish Fisheries 1: 173-205. Tanaka, M., T. Goto, M. Tomiyama, and H. Sudo. 1989. Immigration, settlement and mortality of flounder Paralichthys olivaceus larvae and juveniles in a nursery ground, Shijiki Bay, Japan. Neth. J. Sea Res. 24:57-67. Tominaga, O., and Y. Watanabe. 1998. Geographical dispersal and optimum release size of hatchery-reared Japanese flounder Paralichthys olivaceus released in Ishikari Bay. Hokkaido, Japan. J. Sea Res. 40: 73-81. Travis. J., F. C. Coleman, C. B. Grimes, D. Conover, T. M. Bert, and M. Tringali. 1998. Critically assessing stock enhancement: an introduc- tion to the Mote Symposium. Bull. Mar. Sci. 62: 305-311. van der Veer, H. W., and J. I. Witte. 1993. The "maximum growth/optimal food condition" hypothesis: A test for 0-group plaice Pleuronectes platessa in the Dutch Wadden Sea. Mar. Ecol. Prog. Ser. 101: 81-90. Waters, E. B. 1996. Sustainable flounder culture and fisheries. NC Sea Grant Publication UNC-SG-96-14, 12 p. Sea Grant, Raleigh, NC. Waters. E. B.. and K. Mosher. eds. 1999. Flounder aquaculture and stock enhancement in North Carolina: issues, opportunities and recommendations. NC Sea Grant Publication, UNC-SG-99-02, 24 p. Sea Grant, Raleigh, NC. Willis, S. A.. W. W Falls, C. W. Dennis, D. E. Roberts, and P. G. Whitechurch. 1995. Assessment of season of release and size-at-release on recapture rates of hatchery-reared red drum. Uses and effects of cultured fishes in aquatic ecosystems. Am. Fish. Soc. Symp. 15:354-365. Yamashita, Y, S. Nagahora, H. Yamada, and D. Kitigawa. 1994. Effects of release size on survival and growth of Japa- nese flounder Paralichthys olivaceus in coastal waters off Iwate Prefecture, northeastern Japan. Mar. Ecol. Prog. Ser. 105:269-276. 94 Abstract— Sex-specific demography and reproductive biology- of stripey bass [Lutjanus carponotatus l I also known as Spanish flag snapper. FAO ) were exam- ined at the Palm and Lizard island groups, Great Barrier Reef ( GBR). Total mortality rates were similar between the sexes. Males had larger L . at both island groups and Lizard Island group fish had larger overall L_,, Female:male sex ratios were 1.3 and 1.1 at the Palm and Lizard island groups, respectively. The former is statistically different from 1, but is unlikely significantly different in a biological sense. Females matured on average at 2 years of age and 190 mm fork length at both loca- tions. Female gonadal lipid body indices peaked from August through October, preceding peak gonadosomatic indices in October, November, and December that were twice as great as in any other month. However, ovarian stag- ing revealed 50^ or more ovaries were ripe from September through February, suggesting a more protracted spawning season and highlighting the different interpretations that can arise between gonad weight and gonad staging meth- ods. Gonadosomatic index increases slightly with body size and larger fish have a longer average spawning season, which suggests that larger fish produce greater relative reproductive output. Lizard Island group females had ovaries nearly twice as large as Palm Island group females at a given body size. However, it is unclear whether this reflects spatial differences akin to those observed in growth or effects of sampling Lizard Island group fish closer to their date of spawning. These results support an existing 250 mm minimum size limit for L. carponotatus on the GBR, as well as the timing of a proposed October through December spawning closure for the fishery. The results also caution against assessing reef-fish stocks without reference to sex-, size-, and location-specific biologi- cal traits. Sex-specific growth and mortality, spawning season, and female maturation of the stripey bass {Lutjanus carponotatus) on the Great Barrier Reef Jacob P. Kritzer School of Marine Biology & Aquaculture and CRC Reef Research Centre-Effects of Line Fishing Project James Cook University Townsville. Queensland 4811, Australia Present address: Department of Biological Sciences University of Windsor 401 Sunset Avenue Windsor, Ontario N9B 3P4, Canada E-mail address kntzenSuwindsorca Manuscript approved for publication 22 July 2003 by Scientific Editor. Manuscript received 22 July 2003 at NMFS Scientific Publications Office. Fish Bull. 102:94-107 (2004). Lutjanid snappers are among the most prominent species comprising the catch of hook-and-line fisheries on tropical reefs worldwide (Dalzell, 1996). A notable exception is the line fishery on Australia's Great Barrier Reef (GBR). There, the finfish catch, and therefore the majority of fisheries research, is dominated by coral trouts of the genus Plectropomus (Mapstone et al. 1 ). However, the GBR finfish harvest is diverse and the catch of many sec- ondary species has risen steadily since the early 1990s (Mapstone et al. 1 ). Furthermore, over the past decade, the GBR fishery has changed with the advent of the lucrative Asian live reef- fish market. At present, only a handful of the many species harvested on the GBR are exported to the live reef-fish market. However, continued expansion of the trade coupled with the depletion of fish stocks in other source nations (Bentley 2 ) has the potential to intro- duce demand for a wider range of spe- cies. Even in the absence of changes in the species composition of live reef-fish exports, increased demand for second- ary species due to changes in either domestic preferences or availability of primary species has the potential to elevate harvest of currently nontarget species (Kritzer, 2003). Effective multispecies management of the GBR fishery will ultimately re- quire understanding the biology of more than simply the primary target species. For example, spawning closures of the fishery have been proposed for nine-day periods around the new moon in Octo- ber, November, and December on the rationale that this will protect spawn- ing activity of a wide range of harvested species (Queensland Fisheries Manage- ment Authority 3 ). Yet, spawning season information for species beyond the com- mon coral trout {P. leopardus ) ( Ferreira, 1995; Samoilys. 1997 ) is nearly nonexis- tent. The GBR fishery is in a fortunate position with respect to management of many species for which exploitation is still at relatively low levels because baseline biological characteristics can be estimated before stock structure is drastically altered by fishing. These da- ta can then be used in both formulating management strategies and monitoring effects of fishing. 1 Mapstone. B. D.. J. P. MacKinlay, and C. R. Davies. 1996. A description of the com- mercial reef line fishery log book data held by the Queensland Fisheries Management Authority. Report to the Queensland Fisheries Management Authority. 480 p. Primary Industries Building, GPO Box 4(i. Brisbane. Queensland 4001. Australia. 2 Bentley. N. 1999. Fishing for solutions: can the live trade in wild groupers and wrasses from Southeast Asia be managed? TRAFFIC Southeast Asia report. 143 p. Unit 9-3A, 3rd Floor. Jalan SS23/11, Taman SEA. 47400 Petaling Java, Selan- gor, Malaysia. 3 Queensland Fisheries Management Auth- ority. 1999. Queensland coral reef fin fish fishery. Draft management plan and regulatory impact statement, 80 p. Pri- mary Industries Building. GPO Box 46, Brisbane, Queensland 4001, Australia. Kritzer: Sex-specific growth and mortality, spawning season, and female maturation of Lut/anus carponotatus 95 One of the most prominent secondary species in the GBR fishery is the stripey bass (Lutjanus carponotatus) (Spanish flag snapper. FAO). In relation to other large predators on the GBR, L. carponotatus is highly abundant on inshore reefs, common on mid-continental shelf reefs, and absent from outer-shelf reefs (Newman and Williams, 1996; Newman et al., 1997; Mapstone et al. 4 ). Although this affinity for inshore reefs has the potential to make the species more susceptible to recreational fishing, the limited available data do not suggest that it is heavily exploited by the recreational fleet (Higgs, 1993) in relation to the commercial fleet (Mapstone et al. 1 ). Lutjanus carponota- tus has a broad-based diet, consuming a wide variety of smaller reef fishes and invertebrates (Connell, 1998). Its role as a predator coupled with its abundance, particularly on inshore reefs, suggests that the species might have an important ecological function on the GBR in addition to its role as a fishery resource. Davies (1995) and Newman et al. (2000) have collected basic demographic data for L. carponotatus on the north- ern and central GBR, respectively. They both reported a pronounced asymptote in the growth trajectory and that most growth occurred over the first three to five years and little subsequent growth over a lifespan that can reach 15 to 20 years. Newman et al. (2000) also reported a heavily male-biased sample and larger body sizes among males. Unlike age and growth data, no information on reproduc- tion of L. carponotatus has been available despite that fact that existing (minimum size limits) and proposed (spawn- ing closures) fisheries regulations are based largely on reproductive traits (Queensland Fisheries Management Authority 3 ). Specific aims of this study were 1) to estimate sex ra- tios and sex-specific schedules of growth and mortality; 2) to estimate age- and size-specific schedules of female maturation; 3) to identify the spawning season; and 4) to determine whether reproductive output is proportional to body size by examining the ovary weight-body weight relationship and the average spawning duration of large and small fish. All traits were estimated at the Palm Island group on the central GBR. Additionally, sex-specific growth and female maturity schedules were also examined at the Lizard Island group on the northern GBR to develop spa- tial comparisons. Materials and methods Field methods Size, age, and reproductive data were obtained for 465 L. carponotatus collected by spear fishing on fringing reef slopes during monthly fishery independent sampling at 4 Mapstone, B. D., A.M. Ayling, and J. H.Choat. 1998. Habitat, cross shelf and regional patterns in the distributions and abun- dances of some coral reef organisms on the northern Great Bar- rier Reef. Great Barrier Reef Marine Park Authority research publication 48, 71 p. GPO Box 1379, Townsville, Queensland 4810, Australia. Pelorus, Orpheus, and Fantome Islands in the Palm Island group on the central GBR ( Fig. 1 ) from April 1997 through March 1998. No sampling took place in January 1998 because of severe flooding in the area. To develop spatial comparisons, samples of 118 and 18 fish were obtained in October 1997 and April 1999, respectively, by spear fishing at the Lizard Island group approximately 400 km north of the Palm Island group (Fig. 1). Fish were collected from depths of 2 to 15 m by teams of two to four scuba divers. Lutjanus carponotatus most commonly inhabits depths less than 15 m (Newman and Williams, 1996); therefore sam- pling efforts encountered the majority of the population. Fish were targeted as encountered, without preference based on size, in order to collect as representative a sample as possible. Fish <150 mm fork length (FL) were rare in the samples because they were infrequently observed on reef slopes (Kritzer, 2002). Therefore, supplemental spear fishing on reef flats targeting smaller fish was conducted at the Palm Island group (n=24) in April and December 1999 and at the Lizard Island group (n=25) in May 1999 to obtain growth data for size classes against which the primary sampling was biased. Total weight (TW, g) and FL (mm) of each specimen were recorded. Ovaries and testes of small lutjanids on the GBR are characterized by a lipid body running along the length of each lobe, akin to that found in tropical acanthurids (Fishelson et al., 1985). Gonads and these associated lipid bodies were removed and preserved in FAAC (formaldehyde 4%, acetic acid 5%, calcium chloride 1.3%). Sagittal otoliths were removed, cleaned, and stored for later analyses. Gonad processing and ovarian staging The lipid body was removed from each ovary or testis after fixation and the weight of the gonad (GW) and lipid body (LW) were measured to the nearest 0.01 g. A gonadoso- matic index (GSI) and lipidsomatic index (LSI; after Lobel, 1989) were calculated for each sample as the percentage of TW represented by GW and LW, respectively. Features of whole fixed ovaries including color, speckling, and surface texture were noted as potential criteria for macroscopic staging after comparison with samples processed histologi- cally. Sex of the April 1999 Lizard Island group samples was determined macroscopically only, and was therefore used in sex-specific growth analyses but not in analysis of maturity. Fish <150 mm FL had undeveloped gonads and sex of these specimens was not determined or assigned a reproductive stage. A subsample of 131 ovaries spanning the range of gonad sizes and external appearances were prepared for histo- logical examination. Samoilys and Roelofs (2000) found that medial gonad sections were adequate for determina- tion of reproductive status. Therefore, a medial section was removed from one gonad lobe, dehydrated, and embedded in paraffin. Embedded ovarian tissues were sectioned at 5 nm and stained with hematoxylin and eosin. Ovaries were staged on the basis of the most advanced oocyte stage pres- ent (West, 1990). Additional features used in histological staging included the presence of brown bodies and atretic 96 Fishery Bulletin 102(1 120° 130° N 4 Australia Great LG Barrier 15° Reef PG Queensland 35° Lizard Island group (LG) J\ Lizard Island <\ V-— V^" ,25km, Palfrey Q . ' ' Island _ Seabird Islet South Island Palm Island group (PG) Pelorus Island Brisk Island )). where L t = FL at age t\ L^= the mean asymptotic FL; A' = the Brody growth coefficient; and t = the age at which fish have theoretical FL of 0. Growth functions were fitted by nonlinear least-squares regression of FL on age by using samples for which sex was determined. Because VBGF parameter estimates can be sensitive to the range of ages and sizes used (see Ferreira and Russ. 1994, for an empirical example), a common t equivalent to the .v-intercept of the early growth estimates was used in all models (see "Results" section). Although the sex-specific sample sizes at the Lizard Island group were smaller (n=65 for females; n=62 for males), VBGF parameter estimates achieved high precision at sample sizes between 50 and 100 (Kritzer et al., 2001); therefore the Lizard Island group data were included in the analy- sis. Growth parameters were compared by plotting 959c confidence regions of the parameters K and L x (Kimura, 1980) for each sex from each location and assessing the degree of overlap. Sex-specific total mortality rates, Z, were estimated by using the age-based catch curve of Ricker (1975) as the slope of a linear regression of natural log-transformed fre- quency on age class. Everhart and Youngs ( 1981 ) proposed that catch curve analysis should exclude age classes with n<5 and Murphy ( 1997) proposed that age structures used in catch curves should be truncated at the first age class with n<5. Alternatively, Kritzer et al. (2001) proposed that a sample should contain an average of at least ten fish per age class irrespective of age class-specific sample sizes. Therefore catch curves were fitted by two different methods for each sex at the Palm Island group. The first catch curve began at the modal age class and stopped before the first age class with n < 5. The second catch curve likewise began at the modal age class but included all age classes that were thereafter represented in the data set. Sex-specific sample sizes for the Lizard Island group were too small by any of these criteria and this location was excluded. Mor- tality estimates for Palm Island group fish were compared between the fitting methods within each sex as well as between sexes by ANCOVA. Reproductive biology Maturation schedules of female fish were estimated for each island group by fitting a logistic model, P, = l/(l + exp(a-W)), where P- = the proportion of mature fish in age or 20-mm size class i; a adjusts the position of the curve along the abscissa; and r determines its steepness. Age- and size-specific maturity functions were used to estimate the mean age, r 50 , and size, L 50 , at which 50% of females are mature at each island group. Monthly mean LSI and GSI values of mature Palm Island group fish were plotted separately for males and females to determine seasonal patterns of energy storage and the peak spawning period of L. carponotatus. The pro- portion of specimens at each mature female reproductive stage in each month was also plotted to examine ovarian development patterns throughout the year and the degree of spawning activity occurring outside of peak months. To examine whether relative reproductive output in- creases with body size, GW and GSI for stage-IV ovaries collected during peak spawning months were regressed against TW. Residual plots were used to assess deviation from a linear relationship and to identify three outliers, which were removed from the regression analysis. Regres- sion slopes were compared between the two island groups by ANCOVA. Also, mean GSI values and the proportion of Palm Island group females with stage-IV ovaries during spawning months were compared between females <230 mm FL and those >230 mm FL to examine whether the duration of spawning varies between size classes (nota bene: 230 mm FL is approximately the mean size of mature Palm Island group females and splits each month's sample approximately in half). Results Ovarian staging Five female reproductive stages were identified through histological analysis (Table 1) and were based largely on the scheme of Samoilys and Roelofs (2000). Ovarian stages I (immature) and II (resting mature) have similar oocyte stages. These can be distinguished by the presence of brown bodies or atretic oocytes, which are typically prod- ucts of prior spawning (e.g. Ha and Kinzie, 1996; Adams et al., 2000) and are usually absent from stage-I ovaries. However, these structures will not necessarily persist in ovaries that have spawned, and in fact were rare among the samples; therefore identification of immature females was based primarily on structural organization of the ovary. Stage-I ovaries typically have a thin ovarian wall and more compacted oocytes, whereas ovaries that have previously spawned tend to have a thicker ovarian wall and a more disorganized arrangement of oocytes (Table 1). Also, there were distinct size differences between stage-I ovaries and other stages. The mean GW of stage-I ovaries was approximately one-third that of stage-II ovaries, and mean GSI was approximately one-half of that at stage II (Table 1), and the distribution of body sizes offish at stage I had much lower minimum, maximum, and modal size 98 Fishery Bulletin 102(1) Table 1 Description of histological and macroscopic features (after fixation in a formaldehyde, acetic acid, calcium chloride solution I of ovarian developmental stages of Lutjanus carponotatus. Stage definitions and descriptions are largely a modification of the scheme proposed by Samoilys and Roelofs (2000). Mean ovary weight (GW) and gonosomatic index (GSI) for the larger Palm Island group sample are provided. Stage Histological features Macroscopic features Inactive I Immature II Resting Active III Ripening IVa Ripe IVb Running ripe Relatively thin ovarian wall; lamellae well packed; only darkly purple staining previ- tellogenic oocyte stages (oogonia and peri- nucleolar stages) present. Relatively thick ovarian wall; spaces be- tween lamellae common; only previtellogenic oocyte stages and possibly brown bodies and few atretic vitellogenic oocytes present. Most advanced oocytes are at yolk globule or migratory nucleus stage; atretic oocytes or brown bodies possibly present. Most advanced oocytes at yolk vesicle stage; atretic oocytes or brown bodies possibly present. Similar to stage IVa but large, irregularly shaped, clear to lightly coloured hydrated oocytes are present. Always even white color over entire surface; smooth surface texture; lobes quite small (typi- cally <2 cm long) and thin (mean GW=0.33 g; meanGSI=0.24^). Even white to cream or tan color over gonad sur- face; surface may be smooth or somewhat convo- luted; small white stage II ovaries are difficult to distinguish from stage I without histology I mean GW=1.01 g; mean GSI=0.43%). Color sometimes white but more often cream to tan; surface is commonly convoluted; difficult to distinguish from stage II without histology (mean GW=1.18 g; mean GSI=0.53% I. Color tan to brown or mustard with opaque speck- les that become larger and more dense as late stage oocytes become more numerous; convoluted surface sometimes with prominent vasculariza- tion (mean GW=4.04 g; mean GSI=1.399S \. External appearance identical to stage IVa and can only be differentiated histologically (no sam- ples found at Palm Island group). classes compared with the distribution of body sizes offish at stage II (Fig. 2). Stage-Ill (ripening) ovaries contain oocytes at the yolk vesicle vesicle stage, which some authors classify as vitel- logenic (e.g. Samoilys and Roelofs, 2000) and others classify as previtellogenic (e.g. West 1990). Like stage-II ovaries, stage-Ill ovaries can, but do not necessarily, contain brown bodies or atretic oocytes as evidence of probable prior spawning. Although the fish might not have spawned pre- viously, stage III is considered to be a mature stage in the present study because the appearance of yolk vesicles is associated with the initial development of the yolk globule and represents advanced development of the oocyte beyond perinucleolar stages (West, 1990). Therefore, the fish is pre- paring for spawning and will soon be part of the mature population if it is not already. Mean age and size of stage-II (4.4. years and 219 mm FL), stage-Ill (5.0 years and 222 mm FL), and stage-IV (6.5 years and 261 mm FL) females were much more similar to one another than they were to stage-I females (1.9 years and 119 mm FL). Moreover, size-frequency distributions of fish at stages II, III, and IV showed considerable overlap and similarity with one another and were all quite distinct from the size-frequency distribution for stage-I females (Fig. 2). This suggests a division between immature fish and those that are spawn- ing or are nearly ready to do so. The pronounced difference in GW and GSI between stage-I and stage-Ill ovaries and similarity in these metrics between stage-II and stage-Ill fish (Table 1) further support this division. Most immature ovaries and all ripe ovaries could be identified macroscopically. Because certain macroscopic fea- tures were common to multiple ovarian stages, additional histological features was required to separate the largest immature from the smallest resting ovaries and all ripening from resting ovaries among the samples remaining after the initial comparison betw-een histological and macroscopic features. Only one ovary with fully hydrated oocytes, col- lected at the Lizard Island group, was found among the samples prepared for histological analysis; therefore stages IVa and IVb were treated as a single stage. Stage IV suf- ficiently represents final development toward spawning on the broad seasonal time scale adopted in this study but encompasses a wide range of ovarian characteristics and would need to be divided into more detailed stages for finer temporal scale studies of lunar or diel spawning patterns. No samples exhibited features of truly "spent" ovaries. Sex-specific demography Differences were not apparent in early growth of L. car- ponotatus between the island groups (ANCOVA: df=l, 46; F=1.07; P=0.301); therefore the data were pooled to Kntzer: Sex-specific growth and mortality, spawning season, and female maturation of Lut/anus carponotatus 99 80 70 60 50 H 40 30 H 20 10 rzL estimate an early growth rate of 0.76 mm/d, assuming daily period- icity of micro-increments (Fig. 3). This rate of growth represents quite rapid growth, given that fish are adding 100 mm of length in around 4 months, increasing from approxi- mately 20 to 120 mm FL (Fig. 3). The x-intercept of the early growth curve (=-17.98 d) was divided by 365 d/yr to estimate a common t (=-0.049 yr) for all VBGF models. Although size at age for both sexes at both island groups was character- ized by substantial individual vari- ability, different growth trajectories were evident for males and females (Fig. 4, A and B). Estimates (Table 2) and 95% joint confidence regions (Fig. 4C) for the VBGF parameters indicated that the primary differ- ences in these trajectories at each island group lay in L M (which indi- cated that males grow larger than females). In contrast, the common range of K values spanned by the sexes within each island group indicated similar curvature (Table 2, Fig. 4C). However, use of a common t restricts the range of possible fitted lvalues (Kritzer et al., 2001). In addition to the differences between the sexes, the data revealed a general pattern of larger body sizes at the Lizard Island group (Table 2, Fig. 4). Mortality estimates at the Palm Island group were slightly higher when all age classes beyond 1 year were included compared with exclusion of age classes with n < 5 (Fig. 5). These higher mortality estimates contrast with Murphy's (1997) finding that truncation of the age structure results in higher least-squares estimates of Z. The differ- ences between mortality rates estimated with and without age classes with n < 5 were minor for both males (ANCOVA: df=l, 20; F=0.009; P=0.92) and females (ANCOVA: df=l, 23; F=1.35; P=0.26). Therefore, for comparisons between the sexes, the estimates that included all age classes greater than 1 yr were used. In contrast to the sex- specific growth differences, Z estimates of 0.26/yr and 0.29/yr (Fig. 5) corresponding to annual survivorship of 77% and 75% for females and males, respectively, at the Palm Island group were similar between the sexes (ANCO- VA: df=l, 27; F=0.505; P=0.483). Murphy's (1997) results also suggested that least-squares mortality estimates are likely to be around 30% less than the true mortality rate when n = 200 and the true Z = 0.2/yr. Correcting these mor- tality estimates based upon this potential bias results in Z estimates up to 0.37/yr and 0.41/yr for females and males, respectively, with corresponding annual survivorship of 69% and 66%-. However, the catch curve estimates (Fig. 5) corresponded well with estimates based upon Hoenig's (1983) empirically derived relationship between Z and n In I I □ Stage I Stage II □ Stage III El Stage IV I Bfl 51 1 I i ' ' i ii, mi i m co ,, 3"ir>cor--ooa>o-<-OT- t-t-t-t-^t-t-CMJ(N(MCM and Lizard iBl island groups and estimated 959! joint confidence regions of the parameters A" and l. (C), Parameter estimates are presented in Table 2. Spawning season Mature female LSI values were highest in August through October with a maximum in September ( Fig. 7A). The peak in GSI lagged that of LSI by two months with the high- est values occurring from October through December and with a maximum in November (Fig. 7A>. The absence of a January sample unfortunately leaves some ambiguity as to whether GSI, and therefore presumably spawning activity, would still be high at this time or if it would have begun to decline. Male GSI values also exhibited a November maximum (Fig. 7B). Male LSI values, however, did not show any clear trend of increase and decline throughout the year and peaks in April, May, and August that did not correlate with future GSI values as clearly as seen in the female data (Fig. 7). Unlike LSI values for females, monthly mean male LSI values were always greater than the corresponding GSI values. The seasonal pattern of L. carponotatus spawning activity suggested by monthly trends in the proportions of mature ovarian stages can be interpreted as differ- ent from that suggested by GSI values. The lowest GSI values in the October-December peak period were close to twice as great as the next highest values in Septem- ber and February < Fig. 7A). However, the percentage of stage-IV ovaries in the September sample was greater than 50%. which is well over half the percentage of the October sample; whereas the February sample comprised approximately the same percentage of stage-rV ovaries as October (Fig. 8). Also, more than 50% of the March sample was stage-rV ovaries (Fig. 8). whereas its GSI value was close to that of the months with relatively few ripe ovaries (Fig. 7A). Furthermore, September and March had the highest proportions of ripening (stage-Ill I females and thus far fewer resting mature (stage-II) females than the April to August period of limited spawning activity I Fig. 8). Therefore, regardless of whether September, February, and March are defined as nonspawning months or months of limited spawning activity based upon GSI, analysis of ovarian stage frequencies suggests these to be periods of greater spawning activity than might be predicted with GSI. Clearly, the presence of advanced oocytes is a much better indication of imminent spawning than any measure of gonad size; therefore the reproductive stage- frequency data undoubtedly provide the more accurate picture of L. carponotatus spawning patterns. Of 59 ovaries staged from the October 1997 Lizard Island group sample, eight were at stage I, two were at stage II, and 49 (96% of mature females in the sample) were at stage PV. This finding suggests that the island groups share at least October as a common period of ac- tive spawning. Reproductive differences between locations and among size classes The variation in GW among females of like body sizes during peak spawning months increased to some degree with increasing TW, but there was a generally homoge- neous spread of data around the predicted regression Kntzer: Sex-specific growth and mortality, spawning season, and female maturation of Lut/anus carponotatus 101 male ages 2+: y = -0.289x + 4.319 r 2 = 0.896 male ages with n 0.844 female ages 2+: y = - 0.261 x + 4.557 r 2 = 0.872 4: female ages with n > 4: 0.272X + 4.282 y = - 0.203x + 4.289 0.879 6 8 10 12 Age class (years) 18 Figure 5 Age-based catch curves for female higher elevation lines i and male (♦. lower elevation lines) Lutjanus carponotatus at the Palm Island group fitted to all age classes >1 (solid lines I and age classes >1 with n >4 (dashed lines I. Open symbols represent age class 1, which was not used in the analysis. Table 2 Sex-specific von Bertalanffy growth parameters for Lu tja ms carpom tatus at the Palm and Lizard Island groups , Great Barrier Reef, n is sample size; L F is the mean fork length ( mm ) K is the Brody growth :oefficient per yr) L. is the mean asymptotic fork length (mm); a common t -, of -0.049 yr was used n all growth models. Standarc errors are provided below parameter estimates. n L F A" L. r 2 Palm Island group females 263 224.2 12.11) 0.77 (0.032) 246.3 (2.25) 0.515 males 202 224.7 (2.78) 0.69 (0.028) 264.3 (3.26) 0.629 sex ratio 1.3:1 Lizard Island group females 65 239.9 (4.76) 0.56 (0.043) 263.5 (4.24) 0.618 males 62 256.4 (4.77) 0.51 (0.032) 284.8 (4.03) 0.714 sex ratio 1.1:1 lines across body sizes (Fig. 9A). This suggests that on average GW at stage IV during peak spawning months is a linear function of TW. Lizard Island group fish generally had larger ovaries at a given size than did Palm Island group fish (Fig. 9A), a difference supported by ANCOVA (df=l, 125; F=34.7; P<0.001). In fact, regression slopes of 0.25 and 0.52 suggest relative ovary weights at the Lizard Island group were approximately twice as large as those at the Palm Island group. There were no differences in the GW-TW relationship among October, November, and December at the Palm Island group, and therefore the dif- ferences in this relationship between the island groups was consistent whether only the Palm Island group October data were used or whether the October through December data were used. Although GW is a linear function of TW, the nonzero regression constants (Fig. 9A) mean that GW is not a con- stant proportion of TW. Consequently, GSI increases with increasing TW ( Fig. 9B ). The relationship between TW and GSI is not strong, with regression slopes close to zero and low r 2 values at both island groups (Fig. 9B). Despite this, the relationship is statistically strong at both the Palm ( ANOVA: df=l,82; F=12.70; P=0.006) and Lizard (ANOVA: df= 1,42; F=22.95; P<0.0001) Island groups. Also, there is 102 Fishery Bulletin 102(1) some suggestion that, like the GW-TW relationship, the GSI-TW relationship varies between the island groups, although to a much lesser extent (ANCOVA: df= 1,125; F=7.44;P=0.007). o o A 40 34 22 19 12 16 18 12 4 5 2 1 1 4 1 1.0 - 8 33 6 29 5741 2 1 56 /o E'"B 0.8 - 5/ .' 0.6 - a ■* 0.4 - 15 /"-' 0.2 - 3/ 00 - There is some indication that larger fish spawn over a longer period at the Palm Island group. During the September-February spawning season, mean GSI values were always higher for mature Palm Island group females >230 mm FL compared with mature fe- males <230 mm FL at the same location (Fig. 10). This pattern is likely due in part to the higher relative gonad weights of larger fish (Fig. 9B) but also seems to be driven by greater proportions of stage-IV ovaries among larger mature females in September, October, and February com- pared with fish <230 mm FL (Fig. 10). During these months, 13%, 13% and 25% more large fish were at stage IV, respec- tively, than were small fish. 1 2 3 4 5 6 7 B 9 10 11 12 13 14 15 16 17 M Age class (years) 53 44 25 11 3 2 1.0 0.8 0.6 0.4 0.2 0.0 10 50 90 130 170 210 250 290 330 Size class midpoint (mm fork length) Figure 6 Proportion of mature female Lutjanus carponotatus and estimated age-spe- cific (A) and size-specific (B) logistic maturation schedules at the Palm solid lines) and Lizard (□, broken lines) island groups. Sample sizes for the Palm (top value) and Lizard (lower value) Island groups are presented above the data for each age or size class. Parameters of the maturity functions are provided in Table 3. Discussion Demography and reproduction of L. carponotatus Growth of L. carponotatus is rapid for the first two years of life, slows over the next two years, and nearly ceases by age 4. The slowing and cessation of growth coincide with the ages at 50% and 100% maturity, respectively, and support the argument of Day and Taylor (1997) that maturation represents a pivotal physiological trans- formation and consequently a fundamen- tal shift in the growth trajectory. Further supporting the idea that reproductive development occurs at the expense of somatic growth is the apparently longer average spawning season among larger fish that have ceased most somatic growth. The limited growth over much of the lifes- Table 3 Parameters of age- and size- specific logistic maturation schedules anc estimated ages and fork 1 engths at 50', maturity of female Lu tja n 11 s ca rpon otatus at the Palm and Lizar d Island gi oups, Great Barrier Reef. a adjusts th e posit on of the logi stic function along the abscissa; r determ ines the steepness of the logistic function. f i; n is the age at 50% maturity; L r is the fork length at 509S maturity. Standard errors are provided below parameter estimates. a r r 2 *S0 OI " £ 50 Age-specific Palm Island group 6.40 (1.42) 3.42 (0.12) 0.985 1.9 years Lizard Island group 4.16 (0.48) 1.73 (0.19) 0.990 2.4 years Size-specific Palm Island group 14.72 (1.49) 0.081 (0.008) 0.994 182 mm Lizard Island group 11.61 (3.84) 0.061 (0.020) 0.908 189 mm Kritzer: Sex-specific growth and mortality, spawning season, and female maturation of Lut/anus carponotatus 103 pan of L. carponotatus can explain the apparently constant mortality rate over many age classes (evidenced by high catch curve r 2 values) given that mortality is often largely a function of body size (Roff, 1992). The development and regression of visceral fat stores preceding increases in ovary weight is a pattern that has been observed in other reef fishes, in- cluding tropical surgeonfishes (Acan- thuridae: Fishelson et al., 1985) and groupers (Serranidae: Ferreira, 1995) and temperate rockfishes (Scorpaeni- dae: Guillemot et al., 1985). These pat- terns suggest that the stored lipid is fuelling the energetic costs of spawning. The lack of a similar pattern for males supports the idea that energetic costs associated with production of sperm are low in relation to eggs (Wootton, 1985) thus enabling male L. carponotatus to attain larger sizes, as also reported by Newman et al. (2000). Alternatively, males might spawn more frequently throughout the year than females and the lack of seasonal patterns in lipid storage among males might reflect a more regular energetic demand that precludes energy storage. In any case, these sex-specific growth patterns, coupled with similar mortality rates between the sexes and sex ratios that are at unity or that are at most only slightly female-biased (see below), sug- gest that females are limiting reproduction of this species. Therefore stock dynamics should be modeled in terms of female biology (Hilborn and Walters, 1992). The apparently female-biased sex ratio at the Palm Island group starkly contrasts with the heavily male-biased sex ratio reported for mid-shelf reefs of the central GBR by Newman et al. (2000). However, neither a male- nor fe- male-biased sex ratio would be expected from a nonhermaphrodite that is not known to possess a complex mating system such as defense of females or territories. It is possible that the spawning sex ratio (i.e. excluding juveniles) is closer to unity if males mature earlier than females, but this ratio is not possible to assess because male maturation has not yet been examined for this species. The difference between the sex ratio reported in this study and that by Newman et al. (2000) might be due to variation in mating systems across a cross- shelf density gradient (Newman and Williams, 1996). Alternatively, the sampling by traps and line fishing conducted by Newman et al. (2000) could be more heavily biased toward males than the sampling by spear fishing used in the present study because of larger % 25 to 8 2.0 1.5 1.0 0.5 0.0 Figure 7 Monthly mean gonadosomatic index IGSI ±SE; ■) and lipidsomatic index (LSI ±SE; □) values for mature female (A) and all male (Bl Lutjanus carponotatus at the Palm Island group. B -1.0 P CO - - 0.8 - 0.6 v -H" - 0.4 - 0.2 April June Aug Oct i i Dec Feb Month (1997-98) 100% 80% I" 60% - CD f 40°= 20% 0% li i i I D Stage II Stage III D Stage IV April June Aug Oct Dec Feb Month (1997-98) Figure 8 Monthly frequencies of ovarian stages of mature Lutjanus carpono- tatus at the Palm Island group. Stage descriptions are provided in Table 1. size, wider gape, or more aggressive behavior toward bait among males (Cappo and Brown, 1996). Furthermore, it is likely that a female-biased sex ratio as observed at the 104 Fishery Bulletin 102(1) 30 -i Lizard group: 25 - y = 0.052x -6.33 Ol £ 20 J 1 15 - .-'•" " r 2 = 0.711 5 10 - o 5 - ° -■" . °^^ 200 400 600 800 1000 6.0 - (J 5.0 - B Lizard group: _»--' y = 0.0071X + 0.64 « 4.0- ° a S ° a !-'' f2 = 0353 c g 3.0 - to o 2.0 - c/i o 13 1.0- c <§ 0.0- □ D* u m a £ . ' ° a ° ' ° •*!..■■'' - ^—~—~' n ^4 " D □ m ' • °~ " — m • ' <£J-z i ~*~^° ' Palm group: _*^*~""^° ' . y = 0.0029x + 1.02 ."^1 "■ " r 2 = 0.134 200 400 600 800 1000 Whole body weight (g) Figure 9 Fixed ovary weight (A) and gonadosomatic index iBi at fresh whole body weight for mature female Lutjanus carponotatus at ovarian stage IV (see Table 1) collected during peak spawning months (Oct-Dec) at the Palm solid lines) and Lizard (□, dashed lines) island groups. Sep Oct Nov Dec Jan Month (1997-98) Feb Mar Figure 10 Mean gonadosomatic index (GSI ±SE) for mature female Lutjanus carponotatus at the Palm Island group during the September through March spawning season for small (<230 mm fork length; ■> and large (<230 mm fork length; □) size classes. The percentage of fish at stage IV (see Table 1) is indicated above each data point. Palm Island group is not a prevalent feature of L. carponotatus populations. Rather, the strong statistical suggestion of a sex ratio quite different from unity might be due to the fact that sex ratios often show temporal variability (e.g. Stergiou et al., 1996) coupled with the propensity to achieve statistically significant differences when using large sample sizes (Johnson, 1999). Maturation schedules and sex-specific growth differences were consistent between the island groups, but overall growth pat- terns differed, with Lizard Island group fish reaching larger asymptotic body sizes. Given the vast distance between the island groups, these differences might be due to inherent genetic differences between the populations. Or, effects of temperature (the Palm Island group sits at a higher latitude), turbidity, freshwater run-off (the Palm Island group sits closer to a river mouth and has more developed mangrove systems), or other environmental factors could be driving the differences. Of course, these possibilities are not mutually exclusive. The larger ovaries observed among Liz- ard Island females might be due to further spatial differences or might be an effect of timing of sampling. The temporal resolution of sampling aimed to identify the extent of the spawning season but was too coarse to account for intramonth differences in ovar- ian development. Large changes in ovary size might occur within stage IV, and the final progression to immediate prespawning stages can be rapid (e.g. Davis and West, 1993). The Lizard Island group sample was collected from 17 to 23 October 1998, whereas the corresponding Palm Island group sample was collected from 11 to 12 October 1998. The October 1998 new moon was on the 20 th , and P. leopardus, the only GBR species for which lunar spawning patterns have been reported, spawns primarily around the new moon (Samoilys, 1997). If L. carponotatus spawning is also centered around the new moon, the spatial differences in ovary weight at body weight might be due to more advanced develop- ment toward full hydration within the Lizard Island group sample. In fact, the higher proportion of stage-FV ovaries within the October Lizard Island group sample (96%) compared with the October Palm Island group sample (78'i ), coupled with the higher relative ovary weights at the Lizard Island group in October, can be taken as preliminary evidence that L. carponotatus spawns at the new moon. Comparison with other reef fishes The growth differences between male and female L. carponotatus contrast with a general trend of larger body sizes among female lutjanids observed in Atlan- Kritzer: Sex-specific growth and mortality, spawning season, and female maturation of Lut/anus carponotatus 105 tic, Caribbean, and Hawaiian species (Grimes, 1987). However, the pattern observed in the present study seems common in the Indo-Pacific where males frequently ( Davis and West, 1992; McPherson and Squire, 1992; Newman et al., 1996, 2000). but not universally (Hilomen, 1997), are the larger sex. As noted above, these differences are consis- tent with predictions based on energetic costs of producing sperm and eggs. Lutjanus carponotatus spawning patterns identified by using both GSI and ovarian stage frequencies show pro- nounced seasonal differences: there are at least five months of very limited or no spawning activity from April through August. This finding supports Grimes's ( 1987) observation that continental lutjanid populations tend to have more restricted spawning seasons than populations associated with oceanic islands, which spawn more or less continu- ously throughout the year. Although seasonal patterns ex- ist, the prominence of ripe gonads over seven months from September through March suggests an extended spawning season and supports the general observation that tropical reef fishes spawn over longer periods within the year than do cooler water species (Lowe-McConnell, 1979). However, a study with finer temporal resolution is needed to verify that spawning actually occurs in months with a high pro- portion of stage-IV ovaries. Female L. carponotatus mature on average at approxi- mately 75% of their mean asymptotic size, 54% of their maximum observed size, and 119c of their maximum longevity. The relative size at maturity contrasts with Grimes's ( 1987) observations that shallow-water continen- tal lutjanid populations like those of L. carponotatus on the GBR typically mature at smaller relative sizes (=42% maxi- mum size) compared to deep-water populations associated with oceanic islands (=50% maximum size). Two sympatric shallow -water species, L. russelli (Sheaves, 1995) and L. fulviflamma (Hilomen, 1997), likewise contrast with the general familial trend and mature at approximately 50% and 75% of their maximum size, respectively. Hence, a general pattern of relative size at maturity might exist among shallow-water lutjanids in the GBR region that is different from those regions covered by Grimes's ( 1987 ) review. Lutjanids on the GBR are generally lightly fished (Mapstone et al. 1 ); therefore the geographic difference in sizes at maturity might be due to fishing pressure selecting for smaller sizes at maturity in other regions. The relative age at maturity of L. carponotatus cannot be as readily placed in a broader familial context given that ages at maturity were not widely estimated for lutjanids at the time of Grimes's (1987) review. However, an array of published studies suggests that many tropical and sub- tropical demersal fishes share the absolute, but not relative, ages of L. carponotatus at 50% and 100% maturity at 2 and 4 years, respectively. These include other small gonochores on the GBR (Sheaves, 1995; Hart and Russ, 1996; Hilomen, 1997), as well as a range of gonochores in other regions (Grimes and Huntsman, 1980; Davis and West, 199.3; Ross et al., 1995 ) and hermaphrodites on the GBR and elsewhere (Ferreira, 1993, 1995; Bullock and Murphy, 1994). The ubiquity of this maturity schedule, despite a wide array of maximum body sizes (160-1200 mm) and longevities (6-56 years) among these species, perhaps suggests a common physiological threshold toward which many species gravi- tate in order to maximize lifetime reproductive success. More comprehensive analysis of life history trade-offs (e.g. Roff, 1992) is needed to test this hypothesis. Fisheries management Harvest of L. carponotatus is currently restricted to fish greater than 250 mm total length ( approximately 233 mm FL) with the aim of allowing 50% offish to spawn at least once, and this regulation is proposed to remain after revi- sion by the GBR fishery management plan (Queensland Fisheries Management Authority 3 ). The estimated size at 50% maturity of 190 mm FL suggests that the regula- tion is meeting its objective. However, the objective itself might not adequately protect the reproductive potential of L. carponotatus and similar species if individuals require multiple spawning years to ensure sufficient replenish- ment of the stock. The extensive longevities of many reef fishes have been hypothesized to be a mechanism for coping with low and irregular recruitment rates through a process dubbed the "storage effect" (Warner and Chesson, 1985). The rationale behind the storage effect hypothesis is that fish must reproduce during many breeding seasons in order to endure poor recruitment years and realize high repro- ductive success during the unpredictable and intermittent good recruitment years. If this process is important for population dynamics of L. carponotatus and other species, management will need to protect an intact natural popula- tion structure in some areas within the fishery. Protecting older age classes cannot be achieved by using maximum size limits for species like L. carponotatus that have a pro- nounced asymptote in the growth trajectory because body sizes are similar over a broad range of age classes and size is therefore poorly correlated with age. Protecting natural age structure could be accomplished through a system of strategically designed marine protected areas that allow some populations to experience natural survival free of fishing mortality. Proposed closures of the GBR line fishery during nine- day periods around the new moon in October, November, and December are aimed at protecting spawning activity and particularly spawning aggregations of P. leopardus and other harvested species (Queensland Fisheries Man- agement Authority 3 ). Lutjanus carponotatus shares a peak spawning period during these months with P. leopardus (Ferreira, 1995; Samoilys 1997) and several other sym- patric exploited species (McPherson et al., 1992; Sheaves, 1995; Hilomen, 1997; Brown et al. 5 ). In addition, the larger ovaries of the Lizard Island group fish, which were collected closer to the new moon, may indicate that, like P. leopardus (Samoilys, 1997), L. carponotatus spawns at 5 Brown, I. W., P. J. Doherty, B. Ferreira, C. Keenan, G. McPher- son, G. Russ, M. Samoilys, and W. Sumpton. 1994. Growth, reproduction and recruitment of Great Barrier Reef food fish stocks. Final report to the Fisheries Research and Development Corporation, FRDC Project 90/18, Queensland Department of Primary Industries, 154 p. Southern Fisheries Centre, GPO Box 76. Deception Bay, Queensland 4508, Australia. 106 Fishery Bulletin 102(1 the new moon. Therefore, the timing of the proposed spawn- ing closures seems appropriate. However, it is not known whether L. carponotatus aggregate to spawn; therefore the goal of protecting spawning aggregations might not be rel- evant for this species. In fact, the prevalence and ecological importance of spawning aggregations for any species on the GBR is largely unknown; therefore the efficacy of the proposed closures is difficult to predict. Beyond the implications for management regulations, these data have implications for modeling L. carponotatus stock dynamics. In particular, the results suggest that reproductive output by a unit of L. carponotatus biomass cannot be predicted on the basis of that biomass alone. Relative ovary weight increases slightly with increasing body size and there is evidence that larger fish spawn more frequently. The greatest difference in the proportion of ripe ovaries between size classes occurred in February 1998 af- ter severe flooding in January. It is possible that the lower proportion of ripe ovaries among small fish in February was due to stresses caused by changes in salinity or increased run-off and is not a regular trait. However, increased resil- ience to environmental stresses that allows more frequent spawning would also increase the relative reproductive success of large fish. Therefore, a population comprising fewer larger fish is likely to show greater annual egg pro- duction than a population with equivalent biomass that comprises more numerous but smaller fish. Additionally, the sex-specific patterns reported in this study further suggest gross biomass might be an inadequate index of replenishment potential and that female biomass needs to be considered. Therefore, stock structure, in terms of sex ratio and the frequency of size classes, and not simply overall biomass needs to be considered when predicting reproductive potential. Acknowledgments I thank the numerous assistants who participated in fieldwork, as well as Sam Adams and Sue Reilly for assis- tance with histological examinations. The manuscript was greatly improved by comments from Howard Choat, Carl Walters, Tony Fowler, Campbell Davies, Sam Adams, Bruce Mapstone, an anonymous thesis examiner, and two anony- mous reviewers. This work was conducted while the author was supported by an international postgraduate research scholarship from the Commonwealth of Australia and a postgraduate stipend from the CRC Reef Research Centre. Final preparation of the manuscript took place while the author was supported by a postdoctoral fellowship funded jointly by the University of Windsor and the Canadian National Science and Engineering Research Council (col- laborative research opportunity grant no. 227965-00) to Peter Sale and others). Literature cited Adams, S., B. D. Mapstone, G. R. Russ, and C. R. Davies. 2000. Geographic variation in the sex ratio, sex specific size, and age structure of Plectropomus leopardus (Serranidael between reefs open and closed to fishing on the Great Bar- rier Reef. Can. J. Fish. Aquatic Sci. 57: 1448-1458. Bullock, L. H.. and M. D. Murphy. 1994. Aspects of the life history of the yellowmouth grouper. Mycteroperca interstitialis, in the eastern Gulf of Mexico. Bull. Mar. Sci. 55:30-45. Cappo. M.. and I. W. Brown. 1996. Evaluation of sampling methods for reef fish of com- mercial and recreational interest. CRC Reef Research Centre Technical Report 6. 72 p. Cappo, M., P. Eden, S. J. Newman, and S. Robertson. 2000. A new approach to validation of periodicity and timing of opaque zone formation in the otoliths of eleven species of Lutjanus from the central Great Barrier Reef. Fish. Bull. 98:474-488. Connell, S. D. 1998. Patterns of piscivory by resident predatory reef fish at One Tree Reef, Great Barrier Reef. Mar. Freshw. Res. 49:25-30. Dalzell, P. 1996. Catch rates, selectivity and yields of reef fishing. In Reef fisheries (N. V. C. Polunin and C. M. Roberts, eds.), p. 161-192. Chapman and Hall, London. Davies, C. R. 1995. Patterns of movement of three species of coral reef fish on the Great Barrier Reef. Ph.D. diss, 212 p. James Cook University. Queensland, Australia. Davis, T L. 0„ and G. J. West. 1992. Growth and mortality of Lutjanus vittus (Quoy and Gaimard I from the North West Shelf of Australia. Fish. Bull. 90:395-404. 1993. Maturation, reproductive seasonality, fecundity, and spawning frequency in Lutjanus vittus (Quoy and Gaimard i from the North West Shelf of Australia. Fish. Bull. 91: 224-236. Day, T., and P. D.Taylor. 1997. Von Bertalanffy's growth equation should not be used to model age and size at maturity. Am. Nat. 149: 381-393. Everhart, W H., and W. D. Youngs. 1981. Principles of fishery science, 349 p. Cornell Univ. Press, Ithaca, NY. Ferreira. B. P. 1993. Reproduction of the inshore coral trout Plectropo- mus maculatus (Perciformes: Serranidael from the central Great Barrier Reef, Australia. J. Fish Biol. 42:831-844. 1995. Reproduction of the common coral trout Plectropomus leopardus (Serranidae: Epinephelinael from the central and northern Great Barrier Reef. Australia. Bull. Mar. Sci. 56: 653-669. Ferreira. B. P., and G. R. Russ 1994. Age validation and estimation of growth rate of the coral trout. Plectropomus leopardus, (Lacepede 18021 from Lizard Island, northern Great Barrier Reef. Fish. Bull. 92:46-57. Fishelson, L., W. L. Montgomery, and A. A. Myrberg. 1985. A new fat body associated with the gonad of surgeon- fishes (Acanthuridae: Teleosteil. Mar. Biol. 86:109-112. Grimes, C. B. 1987. Reproductive biology of the Lutjanidae: a review. In Tropical snappers and groupers: biology and fisheries management (J. Polovina and S. Ralston., eds.), p. 239-294. Westview Press, London. Grimes. C. B., and G. R. Huntsman. 1980. Reproductive biology of the vermilion snapper. Kritzer: Sex-specific growth and mortality, spawning season, and female maturation of Lut/anus carponotatus 107 Rhomboplites aurorubens, from North Carolina and South Carolina. Fish. Bull. 78:137-146. Guillemot, P. J., R. J. Larson, and W. H. Lenarz. 1985. Seasonal cycles of fat and gonad volume in five species of northern California rockfish (Scorpaenidae: Sebastes). Fish. Bull. 83:299-311. Ha, P. Y, and R. A. Kinzie. 1996. Reproductive biology of Awaous guamensis, an amphi- dromous Hawaiian goby. Environ. Biol. Fish. 45:383-396. Hart, A. M. and G. R. Russ. 1996. Response of herbivorous fishes to crown-of-thorns Acanthaster planci outbreaks. III. Age, growth, mortality and maturity indices of Acanthurus nigrofuscus. Mar. Ecol. Prog. Ser. 136:25-35. Higgs, J. B. 1993. A descriptive analysis of records of the recreational reef-line fishery on the Great Barrier Reef. M.Sc. thesis, 128 p. James Cook Univ., Queensland, Australia. Hilborn. R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dynamics and uncertainty, 570 p. Kluwer Academic, Boston, MA. Hilomen, V. V. 1997. Inter- and intra-habitat movement patterns and population dynamics of small reef fishes of commercial and recreational significance. Ph.D. diss., 277 p. James Cook Univ., Queensland, Australia. Hoenig, J. M. 1983. Empirical use of longevity data to estimate mortality rates. Fish. Bull. 82:898-902. Johnson, D. H. 1999. The insignificance of statistical significance testing. J. Wildlife Manag. 63:763-772. Kimura, D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. Fish. Bull. 77:765-776. Kritzer. J. P. 2002. Variation in the population biology of stripey bass Lutjanus carponotatus within and between two island groups on the Great Barrier Reef. Mar. Ecol. Prog. Ser. 243:191-207. 2003. Biology and management of small snappers on the Great Barrier Reef. In Bridging the gap: a workshop linking student research with fisheries stakeholders (A. J. Williams, D. J. Welch, G. Muldoon. R. Marriott, J. P. Kritzer, and S. Adams eds. 1. p. 62-80. CRC Reef Research Centre, Townsville, Queensland, Australia. Kritzer, J. P., C. R. Davies, and B. D. Mapstone 2001. Characterizing fish populations: effects of sample size and population structure on the precision of demo- graphic parameter estimates. Can. J. Fish. Aquat. Sci. 58: 1557-1568. Lobel. P. S. 1989. Ocean current variability and the spawning season of Hawaiian reef fishes. Env. Biol. Fish. 24:161-171. Lowe-McConnell, R. H. 1979. Ecological aspects of seasonality in fishes of tropical waters. Symp. Zool. Soc. Lond. 44:219-241. McPherson. G. R., and L. Squire 1992. Age and growth of three dominant Lutjanus species of the Great Barrier Reef inter-reef fishery. Asian Fish. Sci. 5:25-36. McPherson, G.R., L. Squire, and J. O'Brien 1992. Reproduction of three dominant Lutjanus species of the Great Barrier Reef inter-reef fishery. Asian Fish. Sci. 5:15-24. Murphy, M. D. 1997. Bias in Chapman-Robson and least-squares estima- tors of mortality rates for steady state populations. Fish. Bull. 95:863-868. Newman, S. J., and D. M. Williams 1996. Variation in reef associated assemblages of the Lut- janidae and Lethrinidae at different distances offshore in the central Great Barrier Reef. Environ. Biol. Fish. 46: 123-138. Newman, S. J., D. M. Williams, and G. R. Russ 1996. Age validation, growth and mortality rates of the tropi- cal snappers (Pisces: Lutjanidae) Lutjanus adetii (Castel- nau, 18731 and L. quinquelineatus iBloch, 1790) from the central Great Barrier Reef, Australia. Mar. Freshw. Res. 47:575-584. 1997. Patterns of zonation of assemblages of the Lutjanidae, Lethrinidae and Serranidae (Epinephilinaei within and among mid-shelf and outer-shelf reefs in the central Great Barrier Reef. Mar. Freshw. Res. 48:119-128. Newman. S. J., M. Cappo, and D. M. Williams 2000. Age, growth and mortality of the stripey, Lutjanus carponotatus (Richardson) and the brown-stripe snapper, L. vitta I Quoy and Gaimard i from the central Great Barrier Reef, Australia. Fish. Res. 48:263-275. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics offish populations, 382 p. Bull. Fish. Res. Board Can. 191. Roff, DA. 1992. The evolution of life histories, 535 p. Chapman and Hall. New York, NY. Ross, J. L., T M. Stevens, and D. S. Vaughan 1995. Age. growth, mortality, and reproductive biology of red drums in North Carolina waters. Trans. Am. Fish. Soc. 124:37-54. Samoilys. M. A. 1997. Periodicity of spawning aggregations of coral trout Plectropomus leopardus (Pisces: Serranidae) on the northern Great Barrier Reef. Mar. Ecol. Prog. Ser. 160: 149-159. Samoilys, M. A., and A. Roelofs. 2000. Defining the reproductive biology of a large serranid, Plectropomus leopardus. CRC Reef Research Centre Tech- nical Report 31. 36 p. Sheaves, M. 1995. Large lutjanid and serranid fishes in tropical estuar- ies: are they adults or juveniles? Mar. Ecol. Prog. Ser. 129: 31-40. Stergiou, K. I., P. Economidis, and A. Sinis. 1996. Sex ratio, spawning season and size at maturity of red bandfish in the western Aegean Sea. J. Fish. Biol. 49: 561-572. Warner, R. R., and P. L. Chesson. 1985. Coexistence mediated by recruitment fluctuations: a field guide to the storage effect. Am. Nat. 125:769-787. West, G. 1990. Methods of assessing ovarian development in fishes: a review. Aust. J. Mar. Freshw. Res. 41:199-222. Wootton, R. J. 1985. Energetics of reproduction. In Fish energetics: new perspectives (P. Tytler and P. Calow, eds.), p. 231-254. Croom Helm, London. 108 Abstract— The increase in harbor seal (Phoca vitulina richardsi) abundance, concurrent with the decrease in sal- monid [Oncorhynehus spp.) and other fish stocks, raises concerns about the potential negative impact of seals on fish populations. Although harbor seals are found in rivers and estuaries, their presence is not necessarily indicative of exclusive or predominant feeding in these systems. We examined the diet of harbor seals in the Umpqua River, Oregon, during 1997 and 1998 to indi- rectly assess whether or not they were feeding in the river. Fish otoliths and other skeletal structures were recov- ered from 651 scats and used to identify seal prey. The use of all diagnostic prey structures, rather than just otoliths, increased our estimates of the number of taxa, the minimum number of indi- viduals and percent frequency of occur- rence C^FO) of prey consumed. The *7 f FO indicated that the most common prey were pleuronectids, Pacific hake (Merluccius produetus), Pacific stag- horn sculpin [Leptocottus armatus), osmerids. and shiner surfperch (Cyma- togaster aggregata ). The majority ( 76%) of prey were fish that inhabit marine waters exclusively and fish found in marine and estuarine areas (e.g. anad- romous spp. ) which would indicate that seals forage predominantly at sea and use the estuary for resting and opportu- nistic feeding. Salmonid remains were encountered in 39 samples (6%); two samples contained identifiable otoliths, which were determined to be from Chi- nook salmon (O. tshawytscha). Because of the complex salmonid composition in the Umpqua River, we used molecular genetic techniques on salmonid bones retrieved from scat to discern species that were rare from those that were abundant. Of the 37 scats with salmo- nid bones but no otoliths, bones were identified genetically as chinook or coho (O. kisutch) salmon, or steelhead trout (O. mykiss) in 90'? of the samples. Examination of the foraging habits of Pacific harbor seal (Phoca vitulina richardsi) to describe their use of the Umpqua River, Oregon, and their predation on salmonids Anthony J. Orr Adria S. Banks Steve Mellman Harriet R. Huber Robert L. DeLong National Marine Mammal Laboratory Alaska Fisheries Science Center, NMFS, NOAA 7600 Sand Point Way NE Seattle, Washington 98115 E-mail address (for A. J. Orr, contact author) tony.orr gnoaa.gov Robin F. Brown Oregon Department of Fish and Wildlife 2040 S E. Marine Science Drive Newport, Oregon 97365 Manuscript approved for publication 9 October 2003 by Scientific Editor. Manuscript received 20 October 2003 at NMFS Scientific Publications Office. Fish. Bull. 102:108-117 (2004). The Pacific harbor seal (Phoca vitulina richardsi) is found along the west coast of North America from the Aleutian Islands, Alaska, to the San Roque Islands. Baja California (King, 1983; Reeves et al., 1992). Before the pas- sage of the Marine Mammal Protection Act (MMPA) of 1972, harbor seals in Oregon were kept at relatively low numbers (fewer than 500 animals in 1968) because of bounties offered by the state and harassment from commercial and sport fishermen (Pearson and Verts, 1970). Since passage of protective leg- islation, harbor seals in Oregon have increased an average of 6^ to 7% annu- ally between 1978 and 1998, although, in recent years, numbers appear to be leveling at about 8000 individuals (Brown and Kohlmann. 1998). The rapid increase in harbor seal numbers has revived fishery-manag- ers' interest in seal diet because of the potential for increased consumption of commercial fish species. In addition, there has been a heightened concern about greater harbor seal abundance in rivers and estuaries during migra- tions of depressed salmonid popula- tions because of the potential negative impact on the recovery of these fishes (NMFS, 1997). Because of the tenuous status of many salmonid (Oncorhyn- ehus spp. I species along the west coast, the National Marine Fisheries Service ( NMFS ) recommended that the United States Congress modify the MMPA to allow lethal removal of seals from river mouths where they may prey on de- pressed salmonid populations (NMFS. 1997 ). Predation of salmonids by harbor seals in Oregon has been documented (Brown, 1980; Harvey. 1987; Brown et al., 1995; Riemer and Brown, 1997; Beach et al. 1 ). The proportion of salmo- nids in the diet of harbor seals varied from 1% to 30'r depending on area, season, and sampling method (NMFS, 1997). Pinniped prey consumption can be determined from direct observations in some systems, if prey is consumed at 1 Beach, R.. A. Geiger. S. Jefferies. S. Treacy, and B. Troutman. 1985. Marine mam- mals and their interactions with fisheries of the Columbia River and adjacent waters, 1980-1982. NWAFC (Northwest Alaska Fisheries Science Center) processed rep. NWAFC 85-04, 316 p. NWAFC, National Marine Fisheries Service, Seattle, WA, 98115. Orr et al.: Foraging habits of Phoca vitulma richardsi in the Umpqua River, Oregon 109 Pacific Ocean A N the surface (Bigg et al., 1990); however, consumption is typically determined by examining scat (fecal) samples. In the past, species-specific sagittal otoliths found in scats were used exclusively to determine the identification of prey taxa. However, because otoliths can be partially or completely digested, or are not present in scats (because the head of the prey was not consumed ), they are not always an adequate representation of di- et. Recently, investigators have begun to use additional structures (e.g. cranial el- ements, vertebrae) recovered from scats to identify prey (e.g. Olesiuk et al., 1990; Cottrell et al., 1996; Riemer and Brown, 1997; Browne et al., 2002; Lance et al. 2 ). These structures usually are more com- mon than otoliths and frequently can be identified to species; however, bones of some species can be identified to family only (e.g. salmonids). Consequently, the National Marine Mammal Laboratory (NMML) collaborated with the Conser- vation Biology Molecular Genetics Laboratory (CBMGL; Northwest Fish- eries Science Center, Seattle, WA) to develop molecular genetic identification of salmonid species (Purcell et al., 2004). Because of the complex salmonid species composition in the Umpqua River, genetic identification was vital to distinguish species that were rare from those that were abundant. The original impetus of this study was to assess the impact of harbor seal predation on the recovery of the Umpqua River sea-run cutthroat trout (O. clarkii) that were listed as endangered under the Endangered Species Act (ESA) during 1996 (Johnson et al., 1999). Umpqua River cutthroat trout were removed from the ESA in 2000 because they were identified to be part of the larger Oregon Coast evolutionary significant unit (U.S. Fish and Wildlife Service, 2000). The present study was continued despite the "delisting" of cutthroat trout because the Umpqua is inhabited year-round by harbor seals that haul out sev- eral kilometers upriver and is, thus, ideal for determining whether the presence of a pinniped species within a sys- tem is indicative of substantial feeding on fish species of concern within that environment. In addition, the Umpqua River contains several other salmonid species whose status is precarious (NMFS, 1997). Therefore, the development of genetic identification techniques was considered valuable for this system, as well as for future foraging studies in which species-specific identification may be desirable but impossible by way of conventional identification methods. Oregon L mpquu River hauiouts 2 Lance, M., A. Orr, S. Riemer, M. Weise, and J. Laake. 2001. Pinniped food habits and prev identification techniques pro- tocol. AFSC Proc. Rep. 2001-04, 36 p. AFSC, NMFS, NOAA. 7600 Sand Point Way NE, Seattle. WA 98115. Figure 1 Map of the lower section of the Umpqua River, Oregon, where scat samples were collected at two haulout sites during 1997 and 1998. The objectives of this study were 1 ) to determine by an examination of diet if harbor seals that haul out in the Umpqua River feed primarily in the river or elsewhere, and 2) to apply genetic techniques to identify salmonid prey species. Materials and methods Study area The Umpqua River, located in southern Oregon ( Fig. 1 ). is a natal river for sea-run cutthroat trout, as well as chinook (O. tshawytscha), coho (O. kisutch) salmon, and steelhead trout (O. mykiss). The Umpqua estuary is also inhabited year-round by approximately 600-1000 harbor seals and has been designated as an area where pinnipeds and sal- monids significantly co-occur (NMFS, 1997). Scat samples for this study were collected from two hauiouts located within 4.8 km of the river's mouth and within 1.6 km of each other (Fig. 1). Scat collection and analysis Samples were collected during two seasons: "spring" (March through June) and "fall" (August to December). "Spring" corresponded to the migration of anadromous cutthroat trout adults and some juveniles to the ocean and "fall" coincided approximately with the freshwater return of spawning anadromous adults. The migratory and spawn- 110 Fishery Bulletin 102(1) Table 1 Collection dates of harbor seal scats and numbers of scats wi th identifiable prey remains, without identifiable remains and without remains from the Umpqua River, Oregon, during 1997 and 1998 Fall and spring periods correspond to timing of cutthi oat trout runs on the Umpqua River. Collection dates With identifiable remains Without dentifiable remains Without remains Total Fall, 1997 16-23 Sep 26 1 2 29 27 Sep-6 Oct 5 3 8 12-24 Oct 31 7 38 31 Oct-lONov 21 6 27 12-25 Nov 36 10 46 Total 119 1 28 148 Spring 1998 24-25 Mar 27 5 2 34 13-15 Apr 59 5 7 71 26-27 Apr 45 4 4 53 13-14 May 41 4 45 27-28 May 12 1 13 11-12 Jun 35 2 1 38 Total 219 16 19 254 Fall 1998 5-6 Aug 142 1 1 144 19-20 Aug 111 1 3 115 6-9 Sep 28 3 3 34 19-21 Sep 13 13 7-8 Oct 19 1 20 Total 313 5 8 326 ing periods of chinook and coho salmon, and steelhead trout also occur during these times. During fall 1997, all harbor seal scats present at the haulouts were collected every other day during the day- time low tide, weather permitting (Table 1). In 1998. bi- weekly attempts were made to pick a minimum of 50 scats during low tides at the haulout sites (Table 1). Scats were collected, placed in individual plastic bags, and frozen for later processing. At the laboratory samples were thawed and rinsed in nested sieves (1.0 mm, 0.71 mm, and 0.5 mm in 1997; 1.4 mm, 1.0 mm, and 0.5 mm in 1998). Fish struc- tures were dried and stored in glass vials and cephalopod remains were stored in vials with 70"* isopropyl or ethyl alcohol. Prey were identified to the lowest possible taxon by using sagittal otoliths, skeletal, and cartilaginous remains from fish and beaks and statoliths from cephalopods. Other in- vertebrate remains were discarded from analysis because of the uncertainty of identifying them as primary or sec- ondary prey. Unknown prey were categorized as "unidenti- fied" and "unidentifiable" (Browne et al., 2002). Items that were categorized as "unidentifiable" were excluded from analyses because they could not be distinguished from prey already identified in the sample. Otoliths, beaks, and diagnostic bones were identified by using an extensive ref- erence collection at the NMML and voucher samples veri- fied by Pacific Identifications (Victoria, British Columbia). After identification, otoliths were separated by side (left, right, or unknown ) and enumerated to determine minimum number of specific prey. Unique diagnostic structures (e.g. quadrates, angulars, basioccipitals, vomers) were used for identification and enumeration offish. Non-unique skeletal structures such as gillrakers and teeth were used to iden- tify but not enumerate taxa (i.e. their presence indicated only a single individual) unless the structures were from different size classes. Vertebrae were treated like other non-unique structures; however, for salmon, if the number of vertebrae reflected more than one individual, then they were used for enumeration. Cephalopod beaks were sepa- rated by side (upper, lower, or unknown) and enumerated to determine number of prey. To discern where harbor seals were feeding, identified prey were categorized as those exclusively found in rivers or estuaries (e.g. gobiids, cyprinids), those found exclu- sively in marine waters (e.g. gadids, mvxinids), and those that could potentially be found in either environment (e.g. anadromous species, osmerids, petromyzontids) by using Eschmeyer et al. (1983). A seal was considered to feed in the river-estuary system if all the prey taxa identified in the scat were definitely or could potentially be found in the system. For example, a sample containing remains of pea- mouth chub iMylocheilus caurinus), threespine stickleback ( Gasterosteus aculeatus ), river lamprey iLampetra ayresii ), and chinook salmon would be classified as a riverine- Orr et al.: Foraging habits of Phoca vitulina richardsi in the Umpqua River, Oregon 111 estuarine species because these prey items could feasibly be consumed in the river. It was assumed that the seal was feeding in the marine environment if a sample contained exclusively marine prey, such as Pacific hagfish (Eptatretus stoutti). Pacific hake (Merluceius productus), and rockfish (Sebastes spp. ). If a scat comprised prey taxa that poten- tially could be found in a riverine-estuarine system or marine waters (e.g. salmonids, osmerids), as well as those found exclusively in marine waters, then it was assumed that the feeding environment was marine or mixed. Salmonid skeletal remains were sent to the CBMGL for species identification. Remains to be analyzed genetically were selected by number or size (or both) to represent dif- ferent species or individuals present in each scat. For ex- ample, if a scat had 95 approximately equal-size vertebrae (a salmonid has approximately 65 vertebrae; Butler, 1990). then at least two vertebrae (potentially representing at least two individuals) were sent for genetic identification. Also, if a sample had a very large gillraker and three small vertebrae, then the gillraker and one vertebra were sent for genetic identification. The size of diagnostic structures was also used to categorize salmon remains as juvenile or adult, when possible. The CBMGL identified salmonid spe- cies by direct sequencing of mitochondrial DNA or analysis of restriction fragment length polymorphism (Purcell et al., 2004). The abundance of prey taxa in harbor seal diet for each period was described by using the minimum number of individuals (MNI) and percent frequency of occurrence (%FO). We compared the effect of including bone on the number of prey consumed by estimating MNI using the greater number of right or left otoliths and then again using all diagnostic skeletal remains. Cephalopod MNI was estimated from the greater number of upper or lower beaks. The % FO of prey taxon i was defined as I°" %FO, x 100, where O ll; = absence (0) or presence (1) of taxon i in scat k\ and s = the total number of scats that contained identifiable prey remains. The presence of taxon ;' in scat k was determined by using otoliths and then again using all structures. To account for variability in diet, point estimates of %FO for a prey taxon were determined during each sampling period and then averaged for each season. Results Scats Over 725 scats were collected during all periods. The number of scats collected with identifiable remains was 119 (99%; n=148) in fall 1997, 219 (93%; ?z=254) in spring 1998, and 313 (98%; n=326) in fall 1998 (Table 1). Of the 651 samples with identifiable prey remains, 605 (93%) con- tained fish bones, 347 (53%) had fish otoliths, 231 (36%) contained remains from cartilaginous fish, and 41 (6% ) had cephalopod beaks. A majority (65% fall 1997, 65% spring 1998, 63% fall 1998) of scats with identifiable remains had one to three prey taxa present and less than 4% contained more than ten taxa. Approximately 40 prey taxa, repre- senting at least 25 families, were identified throughout the study (Tables 2 and 3). For nearly all prey taxa, MNI was greater when all skel- etal remains were identified than when otoliths were used exclusively (Table 2). For several species, such as Pacific hake. Pacific herring (Clupea pallasii), and Pacific sardine {Sardinops sagax), MNI at least tripled when all structures were used for enumeration (Table 2). For most salmonids, cartilaginous fishes, three-spine stickleback, Irish lords (Hemilepidotus spp.), and Pacific mackerel {Scomber ja- ponicus), no otoliths were recovered; therefore other skel- etal elements had to be used for identification (Table 2). For a few prey, such as cyprinids, gobiids, and butter sole (Isopsetta isolepis), only otoliths were recovered (Table 2). Foraging habits The %FO for most prey taxa was greater when all struc- tures were used than when j ust otoliths were used ( Table 3 ). The %FO indicated that the prey most frequently con- sumed were pleuronectids. Pacific hake. Pacific staghorn sculpin {Leptocottus armatus), osmerids, and shiner surf- perch (Cymatogaster aggregata). Prey frequently found in scats included those that were exclusively marine (e.g. Pacific hake, rex sole (Glyptocephalus zachirus), English sole (Parophiys vetulus), and myxinids), and those that occur in both marine and estuarine waters (e.g. Pacific staghorn sculpin. and shiner surfperch (Table 3] ). Only 24% of scats were composed entirely of prey taxa that could be found in riverine-estuarine systems (Fig. 2). Consequently, a majority of the scats contained prey species that were exclusively marine (.v=25.3%) or were a mixture of marine and potentially marine species (x=50.8%\ Fig. 2). Salmonids Salmonid remains were found in only 6% (39/651) of the samples. Five chinook smolts were identified from otoliths in two samples collected during fall 1997; in the remaining 37 samples, salmonid bones were unidentifiable to species with conventional techniques. With the cooperation of CBMGL, we examined 116 salmonid bones using molecular genetic techniques. Species identification was successful for 67% (78/116) of the bones and teeth from 90% (35/39) of the scat samples that contained salmonid structures. In the four samples that remained unidentified, three con- tained only a single salmonid bone that failed to produce any DNA. Most of the other bones where DNA could not be extracted were small or fragmented and highly digested. Seventeen of the samples contained chinook salmon bones (including the two samples with chinook salmon otoliths); 11 contained coho salmon bones, four contained steelhead trout bones, and three contained bones from two salmonid 112 Fishery Bulletin 102(1) Table 2 Minimum number of individuals ( MNI ) offish prey derived from sagittal otoliths and all structures retrieved from harbor seal scats collected at the Umpqua River during 1997 and 1998. s represents the number of scats with identifiable remains, na indicates taxon did not have sagittal otoliths to be used for identification. Fall 1997(s=119i Spring 1998(s=219i Fall 1998(s=313) MNI MNI MNI MNI MNI MNI Family Species otoliths all structures otoliths all structures otoliths all structures Ammodytidae Pacific sand lance 205 208 317 321 3 7 Bothidae Pacific sanddab 12 13 9 9 1 2 Clupeidae American shad 1 2 4 11 1 15 Pacific herring 6 22 3 10 121 345 Pacific sardine 50 235 39 185 Cottidae Pacific staghorn sculpin 44 65 25 48 30 85 unidentified cottid 8 Cyprinidae peamouth chub 1 1 4 4 4 4 Embiotocidae shiner surfperch 104 109 209 274 23 104 Engraulididae northern anchovy 1 3 1 2 Gadidae Pacific hake 1 35 10 44 58 199 Pacific tomcod 9 21 19 52 8 26 Gasterosteidae threespine stickleback 1 Gobiidae unidentified gobiid 2 2 1 1 Hexagrammidae lingcod 1 1 1 Myxinidae Pacific hagfish 20 13 61 Ophidiidae spotted cusk-eel 4 4 2 2 Osmeridae unidentified osmerid 42 54 14 41 105 132 Petromyzontidae Pacific lamprey na 5 na 89 na 41 river lamprey na 2 na 1 na Pholididae saddleback gunnel 3 7 1 3 1 Pleuronectidae English sole 38 41 37 39 75 84 Dover sole 1 4 5 6 27 51 slender sole 1 1 18 24 28 42 butter sole 1 1 15 15 2 2 rex sole 19 44 44 53 96 125 petrale sole 1 1 starry flounder 10 17 8 12 6 31 Rajidae unidentified rajid na 1 na 7 na 4 Scombridae Pacific mackerel 2 3 2 Scorpaenidae Sebastes spp. 15 6 19 2 3 Trichodontidae Pacific sandfish 1 2 3 Zoarcidae unidentified zoarcid 2 2 Salmonidae coho salmon unknown 4 juvenile 1 4 2 adult 1 3 Steelhead or rainbow trou t unknown 2 2 juvenile 1 chinook salmon unknown 5 6 3 juvenile 5 2 5 adult 1 unidentified salmonid unknown 2 1 2 juvenile 1 1 Orr et al.: Foraging habits of Phoca vitulina richardsi in the Umpqua River, Oregon 113 Table 3 Mean percent frequency of occurrence (%FO) of common prey recovered from harbor seal scat samples collected at haulout sites in the Umpqua River, Oregon, during 1997 and 1998. SD indicates standard deviation. Family Species Fall 1997 Spring 1997 Fall 1998 Mean(±SD) Mean(±SD) Mean(±SDl Ammodytidae Pacific sand lance 12.5 ±8.3 12.6 ±8.3 9.1 ±8.9 Bothidae Pacific sanddab 11.4 ±7.5 4.1 ±2.5 3.0 ±3.2 Clupeidae American shad 4.3 ±0.6 13.0 ±2.3 5.3 ±3.1 Pacific herring 16.9 ±13.7 7.3 ±6.9 35.9 ±21.8 Pacific sardine 16.1 ±12.2 17.9 ±9.1 Cottidae Pacific staghorn sculpin 23.9 ±8.5 21.0 ±19.0 11.8 ±4.5 unidentified cottid 16.5 ±20.4 3.2 ±0.7 0.8 ±0.1 Cyprinidae peamouth chub 3.8 2.3 ±0.6 2.8 Embiotocidae shiner surfperch 18.2 ±8.2 23.6 ±19.4 7.0 ±2.9 Engraulididae northern anchovy 5.5 ±3.2 2.1 ±2.0 Gadidae Pacific hake 27.9+9.7 17.0 ±5.7 41.6 ±25.5 Pacific tomcod 15.4 ±7.8 16.1 ±7.0 12.3 ±8.3 Gasterosteidae threespine stickleback 2.8 Gobiidae unidentified gobiid 7.7 1.7 Hexagrammidae lingcod 3.8 0.7 Loliginidae market squid 12.8 ±10.2 3.5 ±1.3 Myxinidae Pacific hagfish 17.5 ±7.9 6.7 ±3.5 16.5 ±9.4 Octopodidae Octopus rubescens 3.8 ±1.4 8.3 ±2.6 8.4 ±7.0 Ophidiidae spotted cusk-eel 0.9 Osmeridae unidentified osmerid 20.8 ±11.3 14.6 ±8.2 19.5 ±10.0 Petromyzontidae Pacific lamprey 7.7 ±8.2 20.5 ±10.1 8.2 ±2.9 river lamprey 5.6 3.7 Pholididae saddleback gunnel 14.7 ±16.9 2.6 ±0.3 5.3 Pleuronectidae English sole 21.9 ±1.7 8.7 ±5.2 17.5 ±12.0 Dover sole 7.4 ±5.9 4.6 ±0.7 13.5 ±13.6 slender sole 11.0 ±7.2 14.9 ±14.9 butter sole 3.8 7.2 ±3.7 1.4 rex sole 27.4 ±12.1 14.2 ±9.6 19.9 ±20.5 petrale sole 0.7 starry flounder 15.8 ±7.4 3.7 ±1.0 5.8 ±1.2 Rajidae unidentified rajid 2.8 5.0 ±1.6 2.8 Scombridae Pacific mackerel 3.8 ±1.4 4.6 ±4.0 0.8 ±0.1 Scorpaenidae Sebastes spp. 15.7 ±8.3 9.1 ±2.6 2.1 Trichodontidae Pacific sandfish 1.7 2.1 unidentifed bothid/ unidentified flatfish 38.5 ±15.9 20.2 ±10.3 14.8 ±2.5 pleui-onectid Zoarcidae unidentified zoarcid 1.4 Salmonidae coho salmon unknown 5.8 ±3.6 juvenile 4.8 3.3 ±2.3 0.7 adult 2.4 6.2 ±6.2 steelhead/rainbow trout unknown 2.7 ±1.4 0.7 juvenile 0.9 adult 0.9 chinook salmon unknown 7.6 ±3.5 0.8 ±0.1 juvenile 4.0 ±1.1 3.4 3.6 ±3.0 adult 4.8 unidentified salmonid(s) unknown 4.3 ±0.6 2.4 0.8 ±0.1 juvenile 4.8 7.7 114 Fishery Bulletin 102(1) species (two with coho and chinook salmon and one with coho salmon and steelhead trout, Table 2). No cutthroat trout were identified with conventional or molecular genetic techniques. Using otoliths and other diagnostic skeletal struc- tures, we enumerated at least 54 individual salmonids in 39 scats (Table 2). All individuals identified as adults I n =5 ) were coho salmon, except one chinook salmon from spring 1997. Individual juveniles identified as steelhead trout (n=l), coho salmon (re=7), chinook salmon («=12), or unidentified salmonids (/2=2) were present during all periods. Because of the difficulty of determining age from size-variable structures such as gillrakers and teeth, most individuals («=27) were designated as "unknown age." Discussion Investigating diet is essential to assessing the role of harbor seals in marine and freshwater ecosystems in order to quantify their interactions with fisheries and determine their impact on the recovery of endangered species. All methods used to investigate diet of seals and other pinnipeds have some limitations (Murie and Lavigne, 1985, 1986; Harvey, 1989). With scats, it is assumed that the relative frequency of prey identified from undigested remains reflects the frequency of prey eaten (Tollit et al., 1997). However, several investigators have determined that this assumption may be seriously biased in several ways (Hawes, 1983; da Silva and Neilson, 1985; Jobling, 1987; Dellinger and Trillmich, 1988; Harvey, 1989; Pierce and Boyle, 1991; Cottrell et al., 1996; Tollit et al., 1997; Bowen, 2000; Orr and Harvey, 2001). No diet study can estimate detrimental or lethal impacts to prey resulting from harassment by pinnipeds. In addition, once a prey is captured, a seal might consume only the soft tissue (especially of larger prey), which would not leave identifiable evidence in scats. Additionally, because skel- etal remains from different prey species pass through the alimentary canal and erode at different rates they may not reflect the true number or proportions of prey consumed (Hawes, 1983; Harvey, 1989; Pierce and Boyle, 1991; Cottrell et al., 1996; Tollit et al., 1997). Therefore, preda- tion estimates determined from scat samples should be regarded as a measure of minimum impact. Although there are complications inherent in the use of scats to describe the diet of seals, scat analysis remains useful because many scats can be collected quickly, with minimum effort and without harm to the animals (Harvey, 1989). Scats Recently, skeletal remains other than otoliths and beaks have begun to be used to identify and enumerate prey of pinnipeds (e.g. Olesiuk et al., 1990; Cottrell et al., 1996; Riemer and Brown, 1997; Browne et al., 2002). There are constraints, however, for using all skeletal elements to identify prey species, including the need for a reference col- lect ion and the extensive training of personnel to identify Fall I9M7 Q nverine-estuanne marine or mixed Scat categorization Figure 2 Mean percentage plus standard deviation (SD) of scats that were classified as "riverine-estuarine" (i.e. samples composed of prey taxa that are exclusively or potentially (e.g. anadromous species, osmerids) found in rivers or estuaries), "marine" (i.e. samples composed exclusively of prey that inhabit marine waters l, and "marine or mixed" (i.e. samples composed of prey taxa exclusively found in marine waters or those that might inhabit marine waters at some stage in their life). digested prey structures (Cottrell et al., 1996). Moreover, there is usually a bias in the recovery and recognition of prey structures from different taxa (Cottrell et al., 1996; Laake et al., 2002). This bias may be a significant problem in estimating relative abundance of prey or biomass con- sumption by harbor seals and is the reason these indices were not considered in this study. Despite these complications, the use of all available structures increased our estimates of prey diversity, MNI, and % FO for most prey taxa. Examination of all diagnostic structures also allowed us to consider a greater sample size because 93% of scats with identifiable remains contained bones, whereas only 53% of scats contained otoliths. Spe- cies not represented by otoliths, such as salmonids (during 1998) and cartilaginous fishes, were detected because all structures were used. In addition, the MNI of important prey such as Pacific hake. Pacific herring, and Pacific sar- dine would have been greatly underestimated had otoliths been used exclusively because the MNI derived by using all structures was at least threefold greater. Although there are complexities associated with estimating MNI from all structures, this method avoids the use of numerical correc- tion factors determined from recovery rates of otoliths fed to captive seals during laboratory experiments (Browne et al., 2002). Results from captive experiments are highly variable between repeated trials for the same individual and among different individuals (Harvey, 1989; Bowen et al., 2000; Orr and Harvey, 2001 1, Foraging habits Harbor seals in the lower Umpqua River consumed prey from over 35 taxa; however, only a few prey taxa were dominant in their diet, as reflected by %FO. Overall, the five most abundant families of prey were Clupeidae, Cot- Orr et al.: Foraging habits of Phoca vitulina nchardsi in the Umpqua River, Oregon 115 tidae, Embiotocidae, Gadidae, and Pleuronectidae. These are similar to those reported in other studies of harbor seal diet in Oregon (Riemer and Brown, 1997; Browne et al., 2002; Riemer et al. 3 - 4 ). It was evident by the presence of prey like Pacific hake. Pacific sardine, hagfish, and various flatfishes that seals fed offshore in pelagic and demersal areas. Harbor seals also consumed prey (e.g. Pacific staghorn sculpin) com- monly found inshore or in estuarine waters. The NMFS recommendations to remove pinnipeds from systems where endangered prey also occur, rely on the assumption that pinnipeds are primarily feeding (on ESA-listed species) in that system. Our study indicated that this was not the case. Although the seals at the Umpqua hauled out several kilometers up river, they foraged primarily at sea. Because of the life histories of many of the prey taxa, our foraging habitat categories must be considered estimations of where the prey might have been consumed. For example, we estimated that 24% of scats contained prey attributable to the riverine-estuarine environment. However, this may actually be an overestimation because some of these spe- cies potentially inhabit the marine environment at some time in their life and may have been consumed there. Ad- ditionally, scats categorized as marine or mixed may reflect that the seal fed solely in the marine environment (because all the taxa can potentially be found in marine waters) or fed at sea and within the river. Nevertheless, these catego- ries are useful for a broad apportioning of foraging habitat. Even though we were able to determine that approximately 76% of the scats contained marine and potentially marine prey taxa, we were unable to assess whether this reflected a seal population with homogeneous or heterogeneous for- aging patterns. In other words, because the scats could not be attributed to a particular individual, we had no way of discerning: 1) whether the entire seal population foraged roughly three-fourths of the time at sea and one-fourth of the time in the river, or 2) whether 76% of the seals fed at sea whereas 24% foraged closer to shore and in the river. This distinction may be important if only a subgroup of seals is feeding in the river and preying on fish that are seasonally abundant in the estuary, such as salmonids. Studies that incorporate radio- or satellite-telemetry or genetic identification of individual prey items in scats may reveal these distinctions in the future. Because the seals haul out almost 5 km upriver and have been observed as far as 32 km upriver, it is clear that 3 Riemer, S. D., R. F. Brown, and M. I. Dhruv. 1999. Monitoring pinniped predation on salmonids in the Alsea and Rogue River estuaries: fall. 1997. //; Pinniped predation on salmonids: pre- liminary reports on field investigations in Washington, Oregon, and California, p. 104-152. Compiled by National Marine Fisheries Service, Northwest Region. [Available from ODFW, 7118 NE Vandenberg Avenue, Corvallis, OR 97330.] 4 Riemer, S. D., R. F. Brown, and M. I. Dhruv. 1999. Monitoring pinniped predation on salmonids in the Alsea and Rogue River estuaries: fall, 1998. In Pinniped predation on salmonids: pre- liminary reports on field investigations in Washington, Oregon, and California, p. 153-188. Compiled by National Marine Fisheries Service, Northwest Region. [Available from ODFW. 7118 NE Vandenberg Avenue, Corvallis, OR 97330.] seals use the river environment. However, the prevalence of marine fish remains in the scat samples indicates that the seals that haul out at the Umpqua River do not feed exclusively in the river. The predominance of marine prey may reflect a foraging strategy in which the effort required to find marine sources of food is offset by the energy gained by exploiting large aggregations of marine schooling fish (e.g. Pacific hake and Pacific sardine). In this scenario, the seals in the Umpqua estuarine-riverine system may depend on marine resources while taking advantage of protected estuarine waters that provide a sheltered place to rest and occasionally feed. Salmonids We used two methods to estimate the number of salmonids eaten by harbor seals: prey remains and genetic analyses of scat samples. Analysis of skeletal remains was of lim- ited value because the majority of salmonid structures recovered from scat samples were bones, which could be identified only to family. This study represents a novel application of genetic techniques to identify salmonid spe- cies from bones found in scats. These techniques allowed us to determine species for a majority of the salmonid samples that would have otherwise remained unidentified because they did not contain otoliths. Salmonid bones or otoliths were found in 6% of the har- bor seal scats collected during our study — a finding that is comparable to the 5% found by Laake et al. (2002) at the Columbia River. However, it is about one-half of what was found by Riemer and Brown ( 13% ; 1997 1 at selected sites in Oregon. Brown et al. (1995) found salmonids in 12% of gastrointestinal tracts of harbors seals taken incidentally by commercial salmon gillnet fishing operations, and Roffe and Mate (1984) observed that salmonids made up 30% of the prey for harbor seals surface feeding in the Rogue Riv- er. Regardless of sampling method, in these studies, most of the salmonids could be identified only to family because few otoliths were recovered and genetic techniques to identify bones to species had not yet been developed. Salmonids are present in the Umpqua River year-round although species and age composition change throughout the year. In this study, most salmonid prey of known age were juveniles; however, we could determine age of only one-half of the individuals. Juveniles are found in the Umpqua River system year-round and may be easier for seals to catch than adults. Alternatively, perhaps seals did not consume many adult skeletal elements because adult salmonids are large fish, which may be ripped apart rather than swallowed whole. Our sampling seasons encompassed at least some por- tion of the migrations of all salmonids, all of which (except cutthroat trout ) were prey of harbor seals. The fact that portions of all migrations were included in the sampling design was noteworthy because there were a large num- ber of seals in the river throughout the year and yet we found no evidence through genetic or otolith identification that seals consumed cutthroat trout in the Umpqua River. The genetic identification tools developed and applied in our collaboration with CBMGL were useful in discerning 116 Fishery Bulletin 102(1) scarce from abundant salmonids. These techniques may be useful in identifying other pinniped prey that lack spe- cies-specific structures and would allow managers to better assess the impact of pinniped predation on threatened or endangered species. Acknowledgments This study was proposed and initiated in collaboration with Joe Scordino. Scat collection and harbor seal counts were conducted by Lawrence Lehman, Kirt Hughes, Mer- rill Gosho, Sharon Melin, and Robert DeLong. The U.S. Coast Guard Umpqua River Station provided boat storage and a location for keeping a chest freezer during the 1997 field season. We would like to thank the Oregon Institute of Marine Biology, Charleston, OR, where the samples col- lected during 1997 were processed. We greatly appreciate the collaboration with Conservation Biology Molecular Genetics Laboratory, which resulted in the identification of our salmon remains based on genetic methods. We would also like to thank Susan Reimer who kindly helped us with difficult identifications, as well as Lawrence Lehman and Jason Griffith for their verification of bone and otolith identifications. We thank Patience Browne, Patrick Gearin, John Jansen, Mark Dhruv, and three anonymous review- ers for providing helpful comments on earlier drafts of this manuscript. Literature cited Bigg, M. A.. G. Ellis, P. Cottrell, and L. Milette. 1990. Predation by harbour seals and sea lions on adult salmon in Comox Harbour and Cowichan Bay, British Columbia. Can. Tech. Rep. Fish. Aquat. Sci. 1769, 31 p. Bowen, W. D. 2000. Reconstruction of pinnipeds diets: accounting for complete digestion of otoliths and cephalopod beaks. Can. J. Fish. Aquat. Sci. 57:898-905. Brown, R. F. 1980. Abundance, movements and feeding habits of the harbor seal, Phoea vitulina, at Netarts Bay, Oregon. M.S. thesis, 69 p. Oregon State Univ., Corvallis, OR. Brown, R. F., S. D. Riemer, and S. Jefferies. 1995. Food of pinnipeds collected during the Columbia River Area Commercial Salmon Gillnet Observation Program, 1991-1994. ODFW (Oregon Dep. Fish Wildlife), Wildlife Diverstiy Program Tech. Rep. 95-6-01, 16 p. Brown, R. F, and S. Kohlmann. 1998. Trends in abundance and current status of the Pacific harbor seal {Phoea vitulina richardsi) in Oregon: 1977-1998. ODFW, Wildlife Diverstiy Program Tech. Rep. 98-6-01, 16 p. Browne, P., J. L. Laake, and R. L. DeLong. 2002. Improving pinniped diet analyses through identifica- tion of multiple skeletal structures in fecal samples. Fish. Bull. 100:423-433. Butler. U. L. 1990. Distinguishing natural from cultural salmonid dep- osits in Pacific Northwest North America. Ph.D. diss., 218 p. Univ. of Washington, Seattle, WA. Cottrell, P. E., A. W. Trites, and E. H. Miller. 1996. Assessing the use of hard parts in faeces to identify harbour seal prey: results of captive-feeding trials. Can. J. Zool. 74:875-880. da Silva, J., and J. Neilson. 1985. Limitations of using otoliths recovered in scats to estimate prey consumption in seals. Can. J. Fish. Aquat. Sci. 42:1439-1442. Dellinger, T, and F. Trillmich. 1988. Estimating diet composition from scat analysis in otariid seals (Otariidae): is it reliable? Can. J. Zool. 66: 1865-1870. Eschmeyer. W. N, E. S. Herald, and H. Hammann. 1983. A field guide to Pacific Coast fishes of North America, 336 p. Houghton Mifflin Co., Boston, MA. Harvey, J. T. 1987. Population dynamics, annual food consumption, move- ments and dive behaviors of harbor seals, Phoca vitulina richardsi, in Oregon. Ph.D. diss., 177 p. Oregon State Univ., Corvallis, OR. 1989. Assessment of errors associated with harbor seal (Phoca vitulina) faecal sampling. J. Zool., Lond. 219: 101-111. Hawes, S. 1983. An evaluation of California sea lion scat samples as indicators of prey importance. M.S. thesis, 50 p. San Francisco State Univ. San Francisco, CA. Jobling, M. 1987. Marine mammal faeces samples as indicators of prey importance — a source of error in bioenergetics studies. Sarsia 72:255-260. Johnson, O, M. Ruckelshaus, W Grant. F. Waknitz, A. Garrett. G. Bryant, K. Neely, and J. Hard. 1999. Status review of coastal cutthroat trout from Washing- ton, Oregon, and California. NOAA Tech. Memo. NMFS- NWFSC-37, 292 p. King, J. 1983. Seals of the world, 240 p. Comstock Publishing Assoc, Cornell Univ. Press, New York, NY. Laake, J. L., P. Browne. R. L. DeLong, and H. R. Huber. 2002. Pinniped diet composition: a comparison of estimation models. Fish. Bull. 100:434-447. Murie, D., and D. Lavigne. 1985. A technique for the recovery of otoliths from stomach contents of piscivorous pinnipeds. J. Wildl. Manag. 49: 910-912. 1986. Interpretation of otoliths in stomach content analy- ses of phocid seals: Quantifying fish consumption. Can. J. Zool. 64:1152-1157. NMFS (National Marine Fisheries Service). 1997. Investigation of scientific information on the impacts of California sea lions and Pacific harbor seals on salmo- nids and on the coastal ecosystem of Washington, Oregon, and California. NOAA Tech. Memo. NMFS-NWFSC-28, 172 p. Olesiuk, P. E, M. A. Bigg, G. M. Ellis. S. J. Crockford, and R. J. Wigen. 1990. An assessment of the feeding habits of harbour seals (Phoca vitulina i in the Strait of Georgia. British Columbia, based on scat analysis. Can. Tech. Rep. Fish. Aquat. Sci. 1730. 135 p. Orr, A. J., and J. T. Harvey. 2001. Quantifying errors associated with using fecal sam- ples to determine the diet of the California sea lion {Zalo- phu* califbrnianus). Can. J. Zool. 79:1080-1087. Orr et al.: Foraging habits of Phoca vitultna richardsi in the Umpqua River, Oregon 117 Pearson, J., and B. Verts. 1970. Abundance and distribution of harbor seals and northern sea lions in Oregon. Murrelet 51:1-5. Pierce, G. J., and P. R. Boyle. 1991. A review of methods for diet analysis in piscivorous marine mammals. Ocean. Mar. Biol. Ann. Rev. 29:409-486. Purcell, M, G. Mackey, E. LaHood, H. Huber, and L. Park. 2004. Molecular methods for the genetic identification of salmonid prey from Pacific harbor seal ( Phoca vitulina richardsi) scat. Fish. Bull. 102:213-220. Reeves, R., B. Stewart, and S. Leatherwood. 1992. The Sierra Club handbook of seals and sirenians, 359 p. Sierra Club Books, San Francisco. CA. Riemer, S. D., and R. F. Brown. 1997. Prey of pinnipeds at selected sites in Oregon identi- fied by scat (fecal) analysis, 1983-1996. ODFW Wildlife Diversity Program Tech. Rep. 97-6-02. 34 p. Roffe, T, and B. Mate. 1984. Abundances and feeding habits of pinnipeds in the Rogue River, Oregon. J. Wildl. Manag. 48:1262-1274. Tollit, D. J., M.J. Steward, P. M. Thompson, G J. Pierce, M. B. Santos, and S. Hughes. 1997. Species and size differences in the digestion of oto- liths and beaks: implications for estimates of pinniped diet composition. Can. J. Fish. Aquat. Sci. 54:105-119. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to remove the Umpqua River cutthroat trout from the list of endangered wildlife. Federal Register: 26 April 2000, 65181:24420-24422. 118 Abstract— Larval development of the sidestriped shrimp ^Pandalopsis dis- par) is described from larvae reared in the laboratory. The species has five zoeal stages and one postlarval stage. Complete larval morphological charac- teristics of the species are described and compared with those of related species of the genus. The number of setae on the margin of the telson in the first and second stages is variable: 11+12, 12+12, or 11+11. Of these, 11+12 pairs are most common. The present study confirms that what was termed the fifth stage in the original study done by Berkeley in 1930 was the sixth stage and that the fifth stage in the Berkeley's study is comparable to the sixth stage that is described in the present study. The sixth stage has a segmented inner fla- gellum of the antennule and fully devel- oped pleopods with setae. The ability to distinguish larval stages of P. dispar from larval stages of other plankton can be important for studies of the effect of climate change on marine communities in the Northeast Pacific and for marine resource management strategies. Larval development of the sidestriped shrimp (.Pandalopsis dispar Rathbun) (Crustacea, Decapoda, Pandalidae) reared in the laboratory Wongyu Park School of Fisheries and Ocean Sciences University of Alaska Fairbanks Juneau, Alaska, 99801-8677 E-mail address: wparkig'uaf edu R. Ian Perry Pacific Biological Station, Fisheries and Oceans Nanaimo, British Columbia, V9R 5K6, Canada Sung Yun Hong Department of Marine Biology Pukyong National University Pusan, 608-737, Korea Manuscipt approved for publication 23 June 2003 by Scientific Editor. Manuscript received 20 October 2003 at NMFS Scientific Publications Office. Fish. Bull. 102:118-126 (2004). Sixteen species of the genus Pandalop- sis have been recognized in the South- western Atlantic and North Pacific Oceans (Komai, 1994; Jensen, 1998; Hanamura et al., 2000). Most members of the genus attain a large body size and are valuable as commercial fishery resources (Holthuis, 1980; Baba et al., 1986). In the North Pacific, P. dispar, P. ampla, P. aleutica, P. longirostris, P. lucidirimicola, and P. spinosior have been reported. Of these, Pandalopsis dispar is an important component of the commercial shrimp fisheries along with several species of the genus Pandalus. Commercial landings of shrimp during 1999 totaled approximately 19 million tonstPSMFC, 1999). Knowledge of the life histories of these species, including the duration and growth of their larvae, is important for stock assessment and management. However, remarkably little is known about their early life histories because most species of the genus live at con- siderable depths. Of the 16 Pandalopsis species, the larvae of only three species have been described partly or com- pletely from plankton samples or from larvae reared in the laboratory. The larvae of Pandalopsis japoni